CN103577802B - Picture transmitter device, method and program, image recognition Verification System and image received device - Google Patents
Picture transmitter device, method and program, image recognition Verification System and image received device Download PDFInfo
- Publication number
- CN103577802B CN103577802B CN201310306018.XA CN201310306018A CN103577802B CN 103577802 B CN103577802 B CN 103577802B CN 201310306018 A CN201310306018 A CN 201310306018A CN 103577802 B CN103577802 B CN 103577802B
- Authority
- CN
- China
- Prior art keywords
- image
- data volume
- authentication processing
- multiple images
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
本发明提供图像发送装置、方法及程序、图像识别认证系统及图像接收装置。图像发送装置(10)具备输入多个连续的图像的图像输入部(12)、削减向图像输入部(12)输入的多个图像的数据量而生成数据量削减图像的数据量削减部(15)、将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置(20)发送的发送部(19)。数据量削减部(15)通过与由图像接收装置(20)进行的识别认证处理对应的削减方法来削减多个图像的数据量。
The invention provides an image sending device, a method and a program, an image recognition authentication system and an image receiving device. The image transmission device (10) includes an image input unit (12) that inputs a plurality of continuous images, and a data amount reduction unit (15) that reduces the data amount of the plurality of images input to the image input unit (12) to generate a data amount reduced image. ), a transmitting unit (19) that transmits the data-amount-reduced image to an image receiving device (20) that performs identification and authentication processing on the data-amount-reduced image. The data amount reduction unit (15) reduces the data amount of the plurality of images by a reduction method corresponding to the identification and authentication process performed by the image receiving device (20).
Description
在本申请中,主张在2012年7月19日向日本提出申请的专利申请编号2012-160453的利益,该申请的内容通过引用而援引于此。In this application, the benefit of Patent Application No. 2012-160453 filed in Japan on Jul. 19, 2012 is claimed, the contents of which are hereby incorporated by reference.
技术领域technical field
本发明涉及一种进行连续输入的图像的识别或认证的图像识别认证系统、及构成该图像识别认证系统的图像发送装置及图像接收装置、以及图像发送方法及图像发送程序。The present invention relates to an image recognition and authentication system for recognizing or authenticating continuously input images, an image sending device and an image receiving device constituting the image recognition and authentication system, an image sending method, and an image sending program.
背景技术Background technique
以往,在由摄像机和服务器构成的图像识别认证系统中,已知有将通过摄像机的拍摄而得到的图像向服务器发送,通过服务器基于图像进行被拍摄体的识别处理及/或认证处理(以下称为“识别认证处理”)的图像识别认证系统。例如有利用监控摄像机来拍摄人物,作为识别认证处理,进行识别对象(脸)的检测、年龄或性别的判定、个人认证等的图像识别认证系统。Conventionally, in an image recognition authentication system composed of a camera and a server, it is known that an image captured by the camera is sent to the server, and the server performs recognition processing and/or authentication processing (hereinafter referred to as is an image recognition authentication system for "identification authentication processing"). For example, there is an image recognition authentication system that uses a surveillance camera to photograph a person, and performs recognition target (face) detection, age or gender judgment, personal authentication, and the like as recognition authentication processing.
以往,利用摄像机来拍摄连续的图像(动态图像),生成这些连续的多个图像,将它们全部向服务器发送,在服务器侧,对于接收到的全部图像(帧)进行了识别认证处理。如此将全部图像向服务器发送而进行识别认证处理时,图像的传送量增多,不仅作用于网络的负荷增大,而且服务器的处理负荷增大,且用于蓄积由服务器接收到的图像的容量增大等,服务器的成本也增大。Conventionally, a camera shoots continuous images (moving images), generates a plurality of these continuous images, and sends all of them to a server. On the server side, all received images (frames) are identified and authenticated. In this way, when all images are sent to the server for identification and authentication processing, the amount of image transmission increases, not only the load on the network increases, but also the processing load on the server increases, and the capacity for storing images received by the server increases. Larger, etc., the cost of the server also increases.
相对于此,提出了一种在摄像机侧对连续的多个图像进行识别对象的检测处理,仅将检测到识别对象的图像向服务器发送的技术。如此,仅限定为检测到识别对象的图像从摄像机向服务器发送,由此能够减少上述的网络负荷及服务器的成本。In contrast, a technique has been proposed in which a recognition target is detected on a plurality of consecutive images on the camera side, and only images in which the recognition target is detected are transmitted to a server. In this way, only the image in which the identification target is detected is transmitted from the camera to the server, thereby reducing the above-mentioned network load and server cost.
而且,作为用于减少网络负荷及服务器的成本的技术,提出了在摄像机侧,对于连续的多个图像进行识别对象的检测处理,仅选择检测到识别对象的图像中的最佳的图像而向服务器发送的技术(例如,专利文献1至4)。由此,能够限制从摄像机侧向服务器发送的图像的个数,因此能够进一步减少上述的网络负荷及服务器的成本。Furthermore, as a technique for reducing the network load and the cost of the server, it is proposed that on the camera side, the recognition target detection process is performed on a plurality of consecutive images, and only the best image among the images in which the recognition target is detected is selected and uploaded to the camera. Server-sent technology (for example, Patent Documents 1 to 4). Accordingly, since the number of images transmitted from the camera side to the server can be limited, the above-mentioned network load and server cost can be further reduced.
【在先技术文献】【Prior technical literature】
【专利文献】【Patent Literature】
【专利文献1】日本专利第3727798号公报[Patent Document 1] Japanese Patent No. 3727798
【专利文献2】日本专利第4208450号公报[Patent Document 2] Japanese Patent No. 4208450
【专利文献3】日本专利第4021873号公报[Patent Document 3] Japanese Patent No. 4021873
【专利文献4】日本专利第4377472号公报[Patent Document 4] Japanese Patent No. 4377472
然而,以往,在摄像机侧被选择而向服务器发送的图像未必是适合于服务器中的识别认证处理的图像,因此从摄像机向服务器发送而进行识别认证处理的图像的个数受到限制,由此存在识别精度下降的问题。However, in the past, the image selected on the camera side and sent to the server may not be suitable for the identification and authentication process in the server. Therefore, the number of images sent from the camera to the server for identification and authentication is limited. The problem of reduced recognition accuracy.
另外,如上述那样依然存在不使识别精度下降的要求且减少网络负荷及服务器的成本这样的要求。In addition, as described above, there is still a need to reduce the network load and the cost of the server without lowering the recognition accuracy.
发明内容Contents of the invention
本发明鉴于上述的问题而作出,目的是在将通过摄像机的拍摄得到的图像向服务器发送并通过服务器基于图像进行被拍摄体的识别认证处理的图像识别认证系统中,减少网络负荷及服务器的成本。The present invention has been made in view of the above-mentioned problems, and its object is to reduce the load on the network and the cost of the server in an image recognition authentication system that transmits an image captured by a camera to a server and performs identification and authentication processing of a subject based on the image through the server. .
本发明另一目的在于提供一种将通过摄像机的拍摄而得到的图像向服务器发送并通过服务器基于图像进行被拍摄体的识别认证处理的图像识别认证系统,其减少网络负荷及服务器的成本,并抑制由此引起的识别精度的下降。Another object of the present invention is to provide an image recognition and authentication system that sends an image captured by a camera to a server and performs identification and authentication processing on a subject based on the image through the server, which reduces network load and server cost, and The reduction in recognition accuracy caused by this is suppressed.
图像发送装置具备:输入多个连续的图像的图像输入部;削减向图像输入部输入的多个图像的数据量而生成数据量削减图像的数据量削减部;以及将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送的发送部,数据量削减部通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image transmission device includes: an image input unit that inputs a plurality of consecutive images; a data amount reduction unit that generates a data amount reduced image by reducing the data amount of the plurality of images input to the image input portion; The transmission unit transmits the image receiving device that performs identification and authentication processing on the reduced image, and the data amount reducing unit reduces the data amount of the plurality of images by a reduction method corresponding to the identification and authentication process performed by the image receiving device.
图像发送方法包括:输入多个连续的图像的图像输入步骤;削减输入的多个图像的数据量而生成数据量削减图像的数据量削减步骤;以及将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送的发送步骤,数据量削减步骤通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image transmission method includes: an image input step of inputting a plurality of continuous images; a data amount reduction step of reducing the data amount of the input plurality of images to generate a data amount reduced image; and converting the data amount reduced image to the data amount reduced image. In the transmission step of the image receiving device of the recognition and authentication process, the data size reduction step reduces the data size of the plurality of images by a reduction method corresponding to the recognition and authentication process performed by the image receiving device.
图像发送程序使计算机作为图像输入部、数据量削减部、发送部发挥功能,该图像输入部输入多个连续的图像,该数据量削减部削减向图像输入部输入的多个图像的数据量而生成数据量削减图像,该发送部将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送,数据量削减部通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image transmission program causes the computer to function as an image input unit that inputs a plurality of consecutive images, a data volume reduction unit that reduces the data volume of a plurality of images input to the image input unit, and a transmission unit. A data volume reduction image is generated, and the transmission unit transmits the data volume reduction image to an image receiving device that performs identification and authentication processing on the data volume reduction image, and the data volume reduction unit uses a reduction method corresponding to the identification and authentication process performed by the image receiving device. Reduce the amount of data for multiple images.
图像识别认证系统由图像发送装置和图像接收装置构成,图像发送装置具备:输入多个连续的图像的图像输入部;削减向图像输入部输入的多个图像的数据量而生成数据量削减图像的数据量削减部;以及将数据量削减图像向图像接收装置发送的发送部,图像接收装置具备:接收由发送部发送的数据量削减图像的接收部;以及对于由接收部接收的数据量削减图像进行识别认证处理的识别认证处理部,数据量削减部通过与由识别认证处理部进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image recognition authentication system is composed of an image sending device and an image receiving device. The image sending device has: an image input unit that inputs a plurality of continuous images; a data volume reduction unit; and a transmission unit that transmits a data volume reduction image to an image receiving device, and the image reception device includes: a reception unit that receives the data volume reduction image transmitted by the transmission unit; and for the data volume reduction image received by the reception unit The identification and authentication processing unit that performs the identification and authentication processing, and the data volume reduction unit reduce the data volume of the plurality of images by a reduction method corresponding to the identification and authentication processing performed by the identification and authentication processing unit.
图像接收装置具备:接收从图像发送装置发送的图像的接收部;对于由接收部接收的图像进行识别认证处理的识别认证处理部;以及将由识别认证处理部进行的识别认证处理的模式信息向图像发送装置发送的发送部。The image receiving device has: a receiving unit that receives an image sent from the image transmitting device; an identification and authentication processing unit that performs identification and authentication processing on the image received by the receiving unit; The sending unit that the sending device sends.
另一方式的图像接收装置具备:接收从图像发送装置发送的图像的接收部;对于由接收部接收的图像进行识别认证处理的识别认证处理部;存储由识别认证处理部进行的识别认证处理的模式信息与用于生成图像的图像生成方法之间的对应关系的存储部;按照设定的模式信息并参照存储部而决定图像生成方法的图像生成控制部;以及将表示由图像生成控制部决定的图像生成方法的信息向图像发送装置发送的发送部。Another aspect of the image receiving device includes: a receiving unit for receiving an image transmitted from the image transmitting device; an identification and authentication processing unit for performing identification and authentication processing on the image received by the receiving unit; a storage unit for the correspondence relationship between the mode information and the image generation method used to generate the image; an image generation control unit for determining the image generation method by referring to the storage unit according to the set mode information; and determining the display by the image generation control unit The transmission unit that transmits the information of the image generation method to the image transmission device.
如以下说明那样,在本发明中存在其他的方式。因此,本发明的公开想要进行本发明的一部分的提供,并没有意图限制在此记述并请求的发明的范围。As described below, there are other aspects in the present invention. Accordingly, the disclosure of the present invention is intended to provide a part of the present invention, and is not intended to limit the scope of the invention described and claimed herein.
【发明效果】【Invention effect】
根据本发明,利用与识别认证处理对应的方法来削减从图像发送装置向图像接收装置发送的图像的数据量,因此能够减少网络负荷及服务器的成本,并且能够抑制这样的数据量的削减造成的识别认证处理的精度的下降。According to the present invention, the data amount of the image transmitted from the image sending device to the image receiving device is reduced by using a method corresponding to the identification and authentication process, so the load on the network and the cost of the server can be reduced, and the loss caused by such a reduction in the amount of data can be suppressed. Decrease in accuracy of recognition authentication processing.
附图说明Description of drawings
图1是表示实施方式的图像发送装置的结构的框图。FIG. 1 is a block diagram showing the configuration of an image transmission device according to an embodiment.
图2是表示实施方式中的图像识别认证系统的结构的框图。FIG. 2 is a block diagram showing the configuration of the image recognition authentication system in the embodiment.
图3是表示实施方式中的图像接收装置的结构的框图。FIG. 3 is a block diagram showing the configuration of the image receiving device in the embodiment.
图4是表示实施方式中的解析结果的显示例的图。FIG. 4 is a diagram showing a display example of analysis results in the embodiment.
图5是说明实施方式中的从图像接收装置对图像发送装置的模式信息的指定的图。5 is a diagram illustrating designation of mode information from the image receiving device to the image transmitting device in the embodiment.
图6是表示实施方式中的分辨率(像素数)与识别认证的精度的关系的例子的坐标图。6 is a graph showing an example of the relationship between the resolution (the number of pixels) and the accuracy of identification and authentication in the embodiment.
图7是表示实施方式中的参数存储部存储的表格的例子的表。7 is a table showing an example of a table stored in a parameter storage unit in the embodiment.
图8是实施方式中的图像发送装置的图像发送方法的流程图。8 is a flowchart of an image transmission method of the image transmission device in the embodiment.
图9是表示实施方式中的带域成分检测部的带域成分检测处理的流程图。FIG. 9 is a flowchart showing a band component detection process performed by a band component detection unit in the embodiment.
图10是表示实施方式中的图像数据量削减部的图像数据量削减处理的流程图。FIG. 10 is a flowchart showing image data amount reduction processing performed by an image data amount reducing unit in the embodiment.
图11是说明实施方式中的从图像接收装置对图像发送装置的参数的指定的图。FIG. 11 is a diagram illustrating designation of parameters from the image receiving device to the image transmitting device in the embodiment.
图12是表示实施方式中的图像识别认证系统的系统结构的另一例的图。FIG. 12 is a diagram showing another example of the system configuration of the image recognition authentication system in the embodiment.
图13是表示实施方式中的图像识别认证系统的系统结构的又一例的图。Fig. 13 is a diagram showing still another example of the system configuration of the image recognition authentication system in the embodiment.
【标号说明】【Description of labels】
100 图像识别认证系统100 image recognition authentication system
10 图像发送装置10 image sending device
11 拍摄部11 Filming department
12 图像输入部12 Image input section
13 对象检测部13 Object Detection Section
14 对象追踪部14 Object Tracking Department
15 数据削减部15 Data Reduction Department
151 最佳拍摄判定部151 Best Shot Judging Section
152 图像数据量削减部152 Image data volume reduction department
153 对象特征解析部153 Object Characteristic Analysis Department
154 带域成分检测部154 Band component detection unit
155 图像选择部155 Image Selection Section
156 必要区域提取部156 Necessary Area Extraction Department
157 必要带域检测部157 Necessary Band Detection Department
158 分辨率转换部158 Resolution conversion unit
159 图像编码部159 Image Coding Department
16 参数存储部16 Parameter storage unit
17 削减控制部17 Cut Control Department
18 接收部18 Receiving Department
19 发送部19 Sending Department
20 图像接收装置20 image receiving device
21 接收部21 Receiving Department
22 图像译码部22 Image decoding unit
23 对象解析部23 Object Analysis Department
24 数据库24 databases
30 结果显示装置30 Result display device
具体实施方式Detailed ways
以下,叙述本发明的详细的说明。以下说明的实施方式只不过是本发明的例子,本发明能够变形为各种方式。因此,以下公开的特定的结构及功能没有限定专利申请的范围。在本说明书中,“识别认证”这一用语作为表示“识别及/或认证”的用语而使用。Hereinafter, a detailed description of the present invention will be described. Embodiments described below are merely examples of the present invention, and the present invention can be modified into various forms. Therefore, the specific structures and functions disclosed below do not limit the scope of the patent application. In this specification, the term "identification and authentication" is used as a term indicating "identification and/or authentication".
实施方式的图像发送装置具备输入多个连续的图像的图像输入部、削减向图像输入部输入的多个图像的数据量而生成数据量削减图像的数据量削减部、将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送的发送部,数据量削减部通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。An image transmission device according to an embodiment includes an image input unit that inputs a plurality of consecutive images, a data size reduction unit that reduces the data volume of a plurality of images input to the image input unit to generate a data volume reduced image, and converts the data volume reduced image to an object. The data size reduction is a transmission unit that transmits images from the image receiving device for identification and authentication processing, and the data size reduction unit reduces the data size of a plurality of images by a reduction method corresponding to the identification and authentication process performed by the image receiving device.
根据该结构,由于通过与识别认证处理对应的方法来削减从图像发送装置向图像接收装置发送的图像的数据量,因此能够减少网络负荷及服务器的成本,并且能够抑制这样的数据量的削减造成的识别认证处理的精度的下降。According to this structure, since the data amount of the image transmitted from the image sending device to the image receiving device is reduced by a method corresponding to the identification and authentication process, it is possible to reduce the load on the network and the cost of the server, and it is possible to suppress the occurrence of such a reduction in the amount of data. The accuracy of the identification and authentication process is reduced.
在上述的图像发送装置中,数据量削减部可以以与由图像接收装置进行的识别认证处理对应的基准,从多个图像中选择一部分图像,从而削减多个图像的数据量。In the image transmitting device described above, the data amount reducing unit may reduce the data amount of the plurality of images by selecting a part of the images from the plurality of images based on a criterion corresponding to the authentication process performed by the image receiving device.
根据该结构,由于从向图像输入部输入的多个图像中选择一部分图像,因此与发送向图像输入部输入的全部图像的情况相比,能够减少网络负荷及服务器的成本,并且以与识别认证处理对应的基准进行这样的图像选择,因此能够抑制限制图像接收装置中的成为识别认证处理的对象的图像的个数造成的识别认证处理的精度的下降。According to this structure, since some images are selected from a plurality of images input to the image input unit, compared with the case of transmitting all the images input to the image input unit, it is possible to reduce the load on the network and the cost of the server. Since such image selection is performed based on the criteria corresponding to the processing, it is possible to suppress a reduction in the accuracy of the recognition and authentication process caused by limiting the number of images to be the target of the recognition and authentication process in the image receiving device.
在上述的图像发送装置中,数据量削减部可以对多个图像或从多个图像中选择的一部分图像的、与由图像接收装置进行的识别认证处理对应的区域进行修剪,从而削减多个图像的数据量。In the image transmitting device described above, the data amount reducing unit may trim a plurality of images or a part of the images selected from the plurality of images corresponding to the identification and authentication process performed by the image receiving device, thereby reducing the plurality of images. amount of data.
根据该结构,对于多个图像或从多个图像中选择的一部分图像进行修剪,因此与发送向图像输入部输入的图像的整体的情况相比,能削减数据量,能够减少网络负荷及服务器的成本,并以与识别认证处理对应的方法进行这样的修剪,因此能够抑制数据量的削减造成的识别认证处理的精度的下降。According to this configuration, a plurality of images or a part of images selected from a plurality of images are trimmed, so compared with the case of transmitting the entire image input to the image input unit, the amount of data can be reduced, and the load on the network and the load on the server can be reduced. Since such pruning is carried out in a method corresponding to the identification and authentication processing, the reduction in the accuracy of the identification and authentication processing due to the reduction of the amount of data can be suppressed.
在上述的图像发送装置中,数据量削减部可以将多个图像或从多个图像中选择的一部分图像的、与由图像接收装置进行的识别认证处理对应的带域成分提取出,从而削减多个图像的数据量。In the above-mentioned image transmitting device, the data amount reducing unit may extract the band components corresponding to the identification and authentication processing performed by the image receiving device from a plurality of images or a part of images selected from a plurality of images, thereby reducing a large number of images. The data volume of an image.
根据该结构,对于多个图像或从多个图像中选择的一部分图像,提取出一部分带域成分,因此与发送向图像输入部输入的图像的整个带域成分的情况相比,能够削减数据量,而能够减少网络负荷及服务器的成本,并且提取出的带域成分是与识别认证处理对应的带域成分,因此能够抑制数据量的削减造成的识别认证处理的精度的下降。According to this configuration, a part of the band components is extracted from a plurality of images or a part of images selected from a plurality of images, so that the amount of data can be reduced compared to the case of transmitting the entire band components of the image input to the image input unit. , and the network load and the cost of the server can be reduced, and the extracted band component is a band component corresponding to the identification and authentication process, so the reduction in the accuracy of the identification and authentication process caused by the reduction of the amount of data can be suppressed.
在上述的图像发送装置中,数据量削减部可以对多个图像或从多个图像中选择的一部分图像进行分辨率转换,而转换成与由图像接收装置进行的识别认证处理对应的分辨率,从而削减多个图像的数据量。In the image transmitting device described above, the data amount reducing unit may perform resolution conversion on a plurality of images or a part of images selected from the plurality of images, and convert the resolution into a resolution corresponding to the identification and authentication process performed by the image receiving device, The amount of data of a plurality of images is thereby reduced.
根据该结构,对多个图像或从多个图像中选择的一部分图像进行分辨率转换(压缩),因此与不对向图像输入部输入的图像进行分辨率转换(压缩)而发送的情况相比,能削减数据量,能够削减网络负荷及服务器的成本,并以与识别认证处理对应的分辨率进行这样的分辨率转换,因此能够抑制数据量的削减造成的识别认证处理的精度的过度的下降。According to this configuration, the resolution conversion (compression) is performed on a plurality of images or a part of images selected from the plurality of images, so compared with the case where the resolution conversion (compression) is not performed on the image input to the image input unit and transmitted, the The amount of data can be reduced, the network load and the cost of the server can be reduced, and such resolution conversion is performed at a resolution corresponding to the identification and authentication process, so that an excessive decrease in the accuracy of the identification and authentication process due to the reduction of the data amount can be suppressed.
在上述的图像发送装置中,数据量削减部可以以与图像接收装置进行的识别认证处理对应的编码率,对多个图像或从多个图像中选择的一部分图像进行编码,从而削减多个图像的数据量。In the image transmitting device described above, the data amount reducing unit may encode a plurality of images or a part of images selected from the plurality of images at a coding rate corresponding to the identification and authentication process performed by the image receiving device, thereby reducing the number of images. amount of data.
根据该结构,对多个图像或从多个图像中选择的一部分图像进行编码,因此与不对向图像输入部输入的图像进行编码而发送的情况相比,能削减数据量,能够减少网络负荷及服务器的成本,并以与识别认证处理对应的编码率进行这样的编码,因此能够抑制数据量的削减造成的识别认证处理的精度的过度的下降。According to this structure, a plurality of images or a part of images selected from a plurality of images are encoded, so compared with the case where the images input to the image input unit are not encoded and transmitted, the amount of data can be reduced, and the network load and network load can be reduced. Since such encoding is performed at a coding rate corresponding to the identification and authentication processing at the cost of the server, it is possible to suppress an excessive decrease in the accuracy of the identification and authentication processing due to reduction in the amount of data.
上述的图像发送装置可以还具备:接收由图像接收装置进行的识别认证处理的模式信息的接收部;存储模式信息与削减方法之间的对应关系的存储部;参照存储部而决定与由接收部接收到的模式信息对应的削减方法的削减控制部,数据量削减部可以按照由削减控制部决定的削减方法来生成数据量削减图像。The above-mentioned image sending device may further include: a receiving unit that receives the pattern information of the identification and authentication process performed by the image receiving device; a storage unit that stores the correspondence between the pattern information and the reduction method; The reduction control unit of the reduction method corresponding to the received mode information, the data reduction unit may generate the data reduction image according to the reduction method determined by the reduction control unit.
根据该结构,通过从图像接收装置接受该识别认证处理的模式信息,而以与该模式信息对应的削减方法削减图像的数据量而生成数据量削减图像,因此能够抑制图像接收装置中的识别认证处理的精度的下降。According to this configuration, by receiving the pattern information of the identification and authentication process from the image receiving device, the data amount of the image is reduced by a reduction method corresponding to the pattern information to generate a data amount reduced image, so that the identification and authentication process in the image receiving device can be suppressed. Handle a drop in precision.
上述的图像发送装置可以还具备接收表示削减方法的信息的接收部,数据量削减部可以按照用由接收部接收到的信息表示的削减方法而生成数据量削减图像。The image transmission device described above may further include a receiving unit that receives information indicating a reduction method, and the data amount reducing unit may generate a data amount reduced image according to the reduction method indicated by the information received by the receiving unit.
根据该结构,通过从图像接收装置接受表示削减方法的信息,而利用该削减方法来削减图像的数据量而生成数据量削减图像,因此能够抑制图像接收装置中的识别认证处理的精度的下降。According to this configuration, by receiving information indicating the reduction method from the image receiving device, and using the reduction method to reduce the data volume of the image to generate a data volume-reduced image, it is possible to suppress a reduction in the accuracy of identification and authentication processing in the image receiving device.
实施方式的图像发送方法包括:输入多个连续的图像的图像输入步骤;削减输入的多个图像的数据量而生成数据量削减图像的数据量削减步骤;以及将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送的发送步骤,数据量削减步骤通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image transmission method of the embodiment includes: an image input step of inputting a plurality of continuous images; a data amount reduction step of reducing the data amount of the input plurality of images to generate a data amount reduced image; and converting the data amount reduced image to a corresponding data amount In the sending step of reducing images sent by the image receiving device for identification and authentication processing, the data amount reduction step reduces the data amount of a plurality of images by a reduction method corresponding to the identification and authentication processing performed by the image receiving device.
根据该结构,由于利用与识别认证处理对应的方法来削减从图像发送装置向图像接收装置发送的图像的数据量,因此能够削减网络负荷及服务器的成本,并且能够抑制这样的数据量的削减造成的识别认证处理的精度的下降。According to this structure, since the data amount of the image transmitted from the image sending device to the image receiving device is reduced by using a method corresponding to the identification and authentication processing, it is possible to reduce the load on the network and the cost of the server, and it is possible to suppress the occurrence of such a reduction in the amount of data. The accuracy of the identification and authentication process is reduced.
实施方式的图像发送程序使计算机作为图像输入部、数据量削减部、发送部发挥功能,该图像输入部输入多个连续的图像,该数据量削减部削减向图像输入部输入的多个图像的数据量而生成数据量削减图像,该发送部将数据量削减图像向对数据量削减图像进行识别认证处理的图像接收装置发送,数据量削减部通过与由图像接收装置进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image transmission program of the embodiment causes the computer to function as an image input unit that inputs a plurality of consecutive images, a data volume reduction unit that reduces the number of images input to the image input unit, and a transmission unit. data volume to generate a data volume reduced image, the transmission unit sends the data volume reduced image to the image receiving device that performs identification and authentication processing on the data volume reduced image, the data volume reduction unit passes the identification and authentication process corresponding to the image receiving device Reduction method to reduce the data volume of multiple images.
根据该结构,由于利用与识别认证处理对应的方法来削减从图像发送装置向图像接收装置发送的图像的数据量,因此能够削减网络负荷及服务器的成本,并且能够抑制这样的数据量的削减造成的识别认证处理的精度的下降。According to this structure, since the data amount of the image transmitted from the image sending device to the image receiving device is reduced by using a method corresponding to the identification and authentication processing, it is possible to reduce the load on the network and the cost of the server, and it is possible to suppress the occurrence of such a reduction in the amount of data. The accuracy of the identification and authentication process is reduced.
实施方式的图像识别认证系统由图像发送装置和图像接收装置构成,图像发送装置具备:输入多个连续的图像的图像输入部;削减向图像输入部输入的多个图像的数据量而生成数据量削减图像的数据量削减部;以及将数据量削减图像向图像接收装置发送的发送部,图像接收装置具备:接收由发送部发送的数据量削减图像的接收部;以及对于由接收部接收的数据量削减图像进行识别认证处理的识别认证处理部,数据量削减部通过与由识别认证处理部进行的识别认证处理对应的削减方法来削减多个图像的数据量。The image recognition authentication system of the embodiment is composed of an image sending device and an image receiving device, the image sending device has: an image input unit that inputs a plurality of continuous images; and a data amount is generated by reducing the data amount of a plurality of images input to the image input unit. a data volume reduction unit for reducing the image; and a transmission unit for transmitting the data volume reduction image to the image receiving device, the image reception device having: a reception unit for receiving the data volume reduction image transmitted by the transmission unit; and for the data received by the reception unit An identification and authentication processing unit that performs identification and authentication processing on the reduced images, and the data amount reduction unit reduces the data volume of the plurality of images by a reduction method corresponding to the identification and authentication process performed by the identification and authentication processing unit.
根据该结构,利用与识别认证处理对应的方法来削减从图像发送装置向图像接收装置发送的图像的数据量,因此能够削减网络负荷及服务器的成本,并且能够抑制这样的数据量的削减造成的识别认证处理的精度的下降。According to this configuration, the data amount of the image transmitted from the image transmitting device to the image receiving device is reduced by using a method corresponding to the identification and authentication process, so it is possible to reduce the network load and the cost of the server, and it is possible to suppress the loss caused by such a reduction in the amount of data. Decrease in accuracy of recognition authentication processing.
实施方式的图像接收装置具备:接收从图像发送装置发送的图像的接收部;对于由接收部接收的图像进行识别认证处理的识别认证处理部;将由识别认证处理部进行的识别认证处理的模式信息向图像发送装置发送的发送部。An image receiving device according to an embodiment includes: a receiving unit for receiving an image transmitted from an image transmitting device; an identification and authentication processing unit for performing identification and authentication processing on an image received by the receiving unit; A sending unit that sends to an image sending device.
根据该结构,对于发送图像的图像发送装置,发送识别认证处理的模式信息,因此在图像发送装置中,能够以与该模式信息对应的图像生成方法来生成图像,由此,能够抑制识别认证处理部中的识别认证处理的精度的下降。According to this configuration, since the mode information of the identification and authentication process is transmitted to the image transmission device that transmits the image, the image can be generated by the image generation method corresponding to the mode information in the image transmission device, thereby suppressing the recognition and authentication process. Decrease in accuracy of identification and authentication processing in the section.
另一实施方式的图像接收装置具备:接收从图像发送装置发送的图像的接收部;对于由接收部接收的图像进行识别认证处理的识别认证处理部;存储由识别认证处理部进行的识别认证处理的模式信息与用于生成图像的图像生成方法之间的对应关系的存储部;按照设定的模式信息,参照存储部而决定图像生成方法的图像生成控制部;将表示由图像生成控制部决定的图像生成方法的信息向图像发送装置发送的发送部。An image receiving device according to another embodiment includes: a receiving unit that receives an image transmitted from the image transmitting device; an identification and authentication processing unit that performs identification and authentication processing on the image received by the receiving unit; and stores the identification and authentication processing performed by the identification and authentication processing unit. A storage unit for the correspondence between the pattern information of the model information and the image generation method used to generate the image; an image generation control unit that determines the image generation method by referring to the storage unit according to the set mode information; and determining the representation by the image generation control unit The transmission unit that transmits the information of the image generation method to the image transmission device.
根据该结构,对于发送图像的图像发送装置,通知与识别认证处理的模式信息对应的图像生成方法,因此在图像发送装置中,能够按照该图像生成方法生成图像,由此,能够抑制识别认证处理部中的识别认证处理的精度的下降。According to this structure, since the image generation method corresponding to the pattern information of the identification and authentication process is notified to the image transmission device that transmits the image, the image generation method can be generated in the image transmission device according to the image generation method, thereby suppressing the recognition and authentication process. Decrease in accuracy of identification and authentication processing in the section.
以下,参照附图,说明本发明的实施方式的图像识别认证系统。图2是表示本发明的实施方式的图像识别认证系统的结构的框图。如图2所示,图像识别认证系统100具备图像发送装置10和图像接收装置20。图像发送装置10与图像接收装置20经由网络NW连接,能够从图像发送装置10向图像接收装置20传送各种数据,而且能够从图像接收装置20向图像发送装置10传送各种数据。尤其是在图像发送装置10中通过拍摄取得图像,将该图像向图像接收装置20发送。在图像接收装置20中,对于由图像发送装置10接收的图像进行对象的识别认证处理。Hereinafter, an image recognition authentication system according to an embodiment of the present invention will be described with reference to the drawings. FIG. 2 is a block diagram showing the configuration of an image recognition authentication system according to an embodiment of the present invention. As shown in FIG. 2 , the image recognition authentication system 100 includes an image transmitting device 10 and an image receiving device 20 . The image transmitting device 10 and the image receiving device 20 are connected via the network NW, and various data can be transmitted from the image transmitting device 10 to the image receiving device 20 and also can be transmitted from the image receiving device 20 to the image transmitting device 10 . In particular, an image is obtained by photographing in the image transmitting device 10 , and the image is transmitted to the image receiving device 20 . In the image receiving device 20 , an object recognition and authentication process is performed on the image received by the image transmitting device 10 .
在图2中,图像发送装置10及图像接收装置20分别以有线的方式与网络NW连接,但图像发送装置10及图像接收装置20也可以以无线的方式与网络NW连接。尤其是在能够使图像发送装置10移动时,图像发送装置10以无线的方式与网络NW连接的结构有利。In FIG. 2 , the image transmitting device 10 and the image receiving device 20 are respectively connected to the network NW by wire, but the image transmitting device 10 and the image receiving device 20 may be connected to the network NW wirelessly. In particular, when the image transmitting device 10 can be moved, the structure in which the image transmitting device 10 is wirelessly connected to the network NW is advantageous.
图像识别认证系统100通过设定后述的模式,而能够使用于各种用途。图像识别认证系统100能够作为监控系统、销售信息收集系统、车辆识别系统等而应用。例如,在监控系统的情况下,图像发送装置10设置在监控对象部位,图像接收装置20设置在从监控对象部位分离的监控中心,进行作为监控对象的人物的脸的识别认证。而且,例如在车辆识别系统的情况下,图像发送装置10设置在设施的停车场的出口,图像接收装置20设置在设施内的管理室并与设置在设施内的停车费用结算机连接,通过进行车辆的号码的识别认证处理,而进行停车场的出口的门的开闭。The image recognition authentication system 100 can be used for various purposes by setting a mode described later. The image recognition authentication system 100 can be applied as a monitoring system, a sales information collection system, a vehicle identification system, and the like. For example, in the case of a monitoring system, the image transmitting device 10 is installed in a monitoring target site, and the image receiving device 20 is installed in a monitoring center separated from the monitoring target site, and performs face recognition and authentication of a person who is a monitoring target. Moreover, for example, in the case of a vehicle recognition system, the image transmitting device 10 is installed at the exit of the parking lot of the facility, and the image receiving device 20 is installed at the management room in the facility and is connected to a parking fee settlement machine installed in the facility. The identification and authentication process of the vehicle number is performed to open and close the exit door of the parking lot.
图1是表示构成本发明的实施方式的图像识别认证系统的图像发送装置的结构的框图。如图1所示,图像发送装置10具备拍摄部11、图像输入部12、对象检测部13、对象追踪部14、数据削减部15、参数存储部16、削减控制部17、接收部18、及发送部19。图像发送装置10作为数字摄像机构成。FIG. 1 is a block diagram showing the configuration of an image transmission device constituting an image recognition authentication system according to an embodiment of the present invention. As shown in FIG. 1 , the image transmission device 10 includes a photographing unit 11, an image input unit 12, an object detection unit 13, an object tracking unit 14, a data reduction unit 15, a parameter storage unit 16, a reduction control unit 17, a receiving unit 18, and sending part 19. The image transmission device 10 is configured as a digital camera.
拍摄部11具备透镜、光圈、摄像元件、及信号处理电路。拍摄部11拍摄作为识别认证处理的对象的被拍摄体而生成图像。拍摄部11拍摄动态图像,由此生成连续的(例如,每秒30帧的)图像。拍摄部11将生成的图像向图像输入部12输出。The imaging unit 11 includes a lens, a diaphragm, an imaging element, and a signal processing circuit. The imaging unit 11 captures a subject to be recognized and authenticated to generate an image. The imaging unit 11 captures moving images to generate continuous (for example, 30 frames per second) images. The imaging unit 11 outputs the generated image to the image input unit 12 .
图像输入部12连续地输入从拍摄部11输出的图像。对象检测部13对于向图像输入部12输入的图像,进行解析而检测规定的识别认证对象(以下简称为“对象”)。对象例如是人物、人物的脸、车辆的号牌、衣服、鞋等。具体的对象由后述的削减控制部17指定。在对象的检测中,可以使用已知的技术(例如,参照美国专利第7483548号公报、专利第3930504号公报等)。在此,美国专利第7483548号公报及日本专利第3930504号公报的内容通过引用而包含于此。The image input unit 12 continuously inputs the images output from the imaging unit 11 . The object detection unit 13 analyzes the image input to the image input unit 12 to detect a predetermined recognition and authentication object (hereinafter simply referred to as “object”). The objects are, for example, a person, a face of a person, a license plate of a vehicle, clothes, shoes, and the like. A specific target is specified by the reduction control unit 17 described later. For object detection, known techniques can be used (for example, refer to US Patent No. 7,483,548, Patent No. 3,930,504, etc.). The contents of US Patent No. 7483548 and Japanese Patent No. 3930504 are hereby incorporated by reference.
对象追踪部14追踪由对象检测部13检测到的对象,将同一对象彼此沿着时序方向连结。在对象的追踪中,可以使用已知的技术(例如,参照日本特开2004-310281号公报、日本专利第3930504号公报、日本专利第4578864号公报等)。在此,日本特开2004-310281号公报、日本专利第3930504号公报、日本专利第4578864号公报的内容通过引用而包含于此。The object tracking unit 14 tracks the objects detected by the object detection unit 13 and connects the same objects to each other in a time-series direction. For object tracking, known techniques can be used (for example, refer to Japanese Patent Application Laid-Open No. 2004-310281, Japanese Patent No. 3930504, Japanese Patent No. 4578864, etc.). Here, the contents of JP-A-2004-310281, JP-A-3930504, and JP-A-4578864 are incorporated herein by reference.
数据削减部15削减输入的多个连续的图像的数据量而生成数据量削减图像。数据削减部15从多个连续的图像选择一部分图像并削减选择的图像自身的数据量,由此削减从图像发送装置10向图像接收装置20发送的图像的数据量。即,数据削减部15不是将输入的多个图像全部向图像接收装置20发送,而仅发送判断为最佳拍摄的图像(削减发送的图像的个数),由此削减从图像发送装置10向图像接收装置20发送的图像的数据量,并且对于该发送的图像,在利用图像发送装置10削减了该图像自身的数据量的基础上向图像接收装置20发送,因此在这一点上能削减从图像发送装置10向图像接收装置20发送的图像的数据量。The data reduction unit 15 reduces the data amount of a plurality of input continuous images to generate a data amount reduced image. The data reduction unit 15 selects a part of images from a plurality of consecutive images and reduces the data volume of the selected images themselves, thereby reducing the data volume of the images transmitted from the image transmission device 10 to the image reception device 20 . That is, the data reduction unit 15 does not transmit all the input images to the image receiving device 20, but only transmits the images judged to be optimally captured (reduces the number of images to be transmitted), thereby reducing the number of images transmitted from the image transmitting device 10 to the image receiving device 20. The data amount of the image transmitted by the image receiving device 20, and for the transmitted image, the data amount of the image itself is reduced by the image transmitting device 10 and transmitted to the image receiving device 20, so it can be reduced from this point. The data volume of the image transmitted from the image transmitting device 10 to the image receiving device 20 .
数据削减部15具备最佳拍摄判定部151作为用于削减发送的图像的个数的结构,且具备图像数据量削减部152作为用于削减发送的图像自身的数据量的结构。最佳拍摄判定部151及图像数据量削减部152分别以与图像接收装置20中的识别认证处理适合的削减方法,来削减发送的图像的数据量。因此,图像发送装置10具备参数存储部16、削减控制部17、及接收部18。The data reduction unit 15 includes an optimal shot determination unit 151 as a configuration for reducing the number of images to be transmitted, and an image data amount reduction unit 152 as a configuration for reducing the data amount of the images themselves to be transmitted. The optimum shot determination unit 151 and the image data size reduction unit 152 reduce the data size of the image to be transmitted by a reduction method suitable for the identification and authentication process in the image receiving device 20 . Therefore, the image transmission device 10 includes a parameter storage unit 16 , a reduction control unit 17 , and a reception unit 18 .
接收部18从图像接收装置20接收识别认证处理的模式信息。该模式信息对应于图像识别认证系统100的上述的各种用途。模式信息与各种削减参数建立对应而存储于在参数存储部16。削减控制部17从参数存储部16读出与由接收部18接收到的模式信息对应的削减参数,设定在对象检测部13及数据削减部15的各种处理中使用的参数。模式信息与削减参数的关系、及使用了削减参数的各种处理,在后面叙述。The receiving unit 18 receives mode information for identification and authentication processing from the image receiving device 20 . This pattern information corresponds to the above-mentioned various uses of the image recognition authentication system 100 . The mode information is stored in the parameter storage unit 16 in association with various reduction parameters. The reduction control unit 17 reads the reduction parameters corresponding to the pattern information received by the reception unit 18 from the parameter storage unit 16 , and sets parameters used in various processes of the object detection unit 13 and the data reduction unit 15 . The relationship between the mode information and the reduction parameters, and various processes using the reduction parameters will be described later.
最佳拍摄判定部151由对象特征解析部153、带域成分检测部154、及图像选择部155构成。对象特征解析部153对于由对象检测部13检测到的对象进行特征解析,算出可靠度(例如脸相似度)、对象物的方向(例如脸的朝向)等。在特征解析中可以使用已知的技术(例如,参照日本特开2005-227957号公报等)。在此,日本特开2005-227957号公报的内容通过引用而包含于此。带域成分检测部154按照由削减控制部17设定的削减参数,检测在通过图像接收装置20进行的识别认证处理中重视的带域成分。The optimal shot determination unit 151 is composed of an object characteristic analysis unit 153 , a band component detection unit 154 , and an image selection unit 155 . The object feature analysis unit 153 performs feature analysis on the object detected by the object detection unit 13 to calculate reliability (for example, face similarity), direction of the object (for example, face orientation), and the like. Known techniques can be used for feature analysis (for example, refer to Japanese Patent Laid-Open No. 2005-227957, etc.). Here, the content of Unexamined-Japanese-Patent No. 2005-227957 is incorporated here by reference. The band component detection unit 154 detects a band component that is important in the identification and authentication process performed by the image receiving device 20 according to the reduction parameter set by the reduction control unit 17 .
图像选择部155基于对象特征解析部153的特征解析的结果、及带域成分检测部154的带域成分检测的结果,从包含同一对象的多个图像中选择最佳拍摄(最佳的图像),向图像数据量削减部152输出(以下,将由最佳拍摄判定部151选择的图像称为“选择图像”)。具体而言,按照由削减控制部17设定的参数,决定对由对象特征解析部153得到的可靠度和方向、及由带域成分检测部154得到的带域成分中的哪一个重视到何种程度,以其为选择基准而选择图像。图像选择部155例如在进行年龄判定作为图像接收装置20中的识别认证处理时,以与脸的方向相比更重视包含多个高频成分的图像这样的基准,选择图像。The image selection unit 155 selects the best shot (best image) from a plurality of images including the same object based on the result of the feature analysis by the object feature analysis unit 153 and the result of the band component detection by the band component detector 154 . , and output to the image data amount reduction unit 152 (hereinafter, the image selected by the optimal shot determination unit 151 is referred to as a “selected image”). Specifically, according to the parameters set by the reduction control unit 17, it is determined which of the reliability and direction obtained by the object feature analysis unit 153 and the band component obtained by the band component detection unit 154 is given more importance. To a certain extent, images are selected using it as a selection criterion. The image selection unit 155 selects an image based on the criterion that more emphasis is placed on an image containing a large number of high-frequency components than the orientation of a face when performing age determination as identification and authentication processing in the image receiving device 20 , for example.
图像数据量削减部152具备必要区域提取部156、必要带域提取部157、分辨率转换部158、及图像编码部159。必要区域提取部156按照由削减控制部17设定的削减参数,从选择图像算出适合于通过图像接收装置20进行识别认证处理的区域而提取(修剪)。必要带域提取部157按照由削减控制部17设定的削减参数,从选择图像算出适合于通过图像接收装置20进行识别认证处理的带域而提取。分辨率转换部158按照通过削减控制部17设定的削减参数,以成为通过图像接收装置20进行识别认证处理所需的最小限度的分辨率(像素数)的方式进行选择图像的分辨率转换(重新设定尺寸)。图像编码部159按照由削减控制部17设定的削减参数,以成为通过图像接收装置20进行识别认证处理所需的最小限度的编码率的方式对选择图像进行编码。The image data amount reduction unit 152 includes a necessary region extraction unit 156 , a necessary band extraction unit 157 , a resolution conversion unit 158 , and an image coding unit 159 . The necessary region extraction unit 156 calculates and extracts (trimmed) a region suitable for recognition and authentication processing by the image receiving device 20 from the selected image according to the reduction parameter set by the reduction control unit 17 . The necessary band extracting unit 157 calculates and extracts a band suitable for the identification and authentication process by the image receiving device 20 from the selected image according to the reduction parameter set by the reduction control unit 17 . The resolution conversion unit 158 converts the resolution of the selected image so that the minimum resolution (the number of pixels) required for the identification and authentication process by the image receiving device 20 is obtained in accordance with the reduction parameter set by the reduction control unit 17 ( resize). The image encoding unit 159 encodes the selected image at the minimum encoding rate required for the identification and authentication process by the image receiving device 20 according to the reduction parameter set by the reduction control unit 17 .
发送部19将从数据削减部15输出的图像和与该图像相关的附属信息向图像接收装置20发送。如上所述,由发送部19向图像接收装置20发送的图像是从由拍摄部11向图像输入部12连续输入的多个图像选择一部分作为选择图像的基础上,对选择图像进行修剪、重新设定尺寸、及编码,由此削减其数据量的图像(数据量削减图像)。如此,在图像发送装置10中,进行从多个图像选择一部分并削减该选择图像自身的数据量的处理(修剪、重新设定尺寸、编码),因此能够削减从图像发送装置10向图像接收装置20发送的图像的数据量,由此,能够减少网络NW的负荷及图像接收装置20的处理负荷、数据库的必要容量。The transmission unit 19 transmits the image output from the data reduction unit 15 and the attached information related to the image to the image reception device 20 . As described above, the image transmitted from the transmitting unit 19 to the image receiving device 20 is selected as a selected image from a plurality of images continuously input from the imaging unit 11 to the image input unit 12, and the selected image is trimmed and reset. An image whose data volume is reduced by sizing and encoding (data volume reduced image). In this way, in the image transmitting device 10, the process of selecting a part from a plurality of images and reducing the data amount of the selected image itself (trimming, resizing, encoding) can be reduced. The amount of image data transmitted by 20 can thereby reduce the load on the network NW, the processing load on the image receiving device 20, and the required capacity of the database.
另一方面,在最佳拍摄判定部151中,选择与图像接收装置20的识别认证处理适合的图像,而且,在图像数据量削减部152的修剪中,提取出图像接收装置20的识别认证处理所需的区域,在图像数据量削减部152的重新设定尺寸及编码中,以图像接收装置20的识别认证处理所需的限度进行数据量的削减。由此,通过基于最佳拍摄判定部151及图像数据量削减部152的数据量的削减,能够抑制图像接收装置20的识别认证处理的精度的下降。On the other hand, in the optimal shot determination unit 151, an image suitable for the identification and authentication process of the image receiving device 20 is selected, and in the pruning of the image data volume reduction unit 152, the image for the identification and authentication process of the image receiving device 20 is extracted. In the necessary area, the image data amount reduction unit 152 resizes and encodes the data amount to the limit necessary for the identification and authentication process of the image receiving device 20 . As a result, the reduction in the accuracy of the identification and authentication processing of the image receiving device 20 can be suppressed by reducing the data size by the optimum shot determination unit 151 and the image data size reduction unit 152 .
图3是表示图像接收装置20的结构的框图。如图3所示,图像接收装置20具备接收部21、图像译码部22、对象解析部23、数据库24、模式设定部25、及发送部26。图像接收装置20构成作为图像识别认证服务器,与作为数字摄像机而构成的图像发送装置10连接。FIG. 3 is a block diagram showing the configuration of the image receiving device 20 . As shown in FIG. 3 , the image receiving device 20 includes a receiving unit 21 , an image decoding unit 22 , an object analyzing unit 23 , a database 24 , a mode setting unit 25 , and a transmitting unit 26 . The image receiving device 20 constitutes an image recognition authentication server, and is connected to the image transmitting device 10 constituted as a digital camera.
接收部21接收从图像发送装置10的发送部19发送的数据量削减图像及该图像的附属信息。图像译码部22对接收到的图像进行译码。对象解析部23对译码后的图像及附属信息进行解析,而进行对象的识别认证处理。在此的识别认证处理例如是从人物的脸的年龄及性别的识别、从人物的脸的特定的人物的认证、车辆的号牌的识别等。数据库24保存由图像译码部22译码的图像及由对象解析部23得到的解析结果。The receiving unit 21 receives the data volume-reduced image and the attached information of the image transmitted from the transmitting unit 19 of the image transmitting device 10 . The image decoding unit 22 decodes the received image. The object analysis unit 23 analyzes the decoded image and attached information, and performs identification and authentication processing of the object. The identification and authentication processing herein includes, for example, identification of the age and gender of a person's face, identification of a specific person from the face of a person, identification of a vehicle's license plate, and the like. The database 24 stores images decoded by the image decoding unit 22 and analysis results obtained by the object analyzing unit 23 .
如图3所示,在图像接收装置20上还连接有结果显示装置30。结果显示装置30例如是液晶显示面板,显示保存于数据库24的图像及/或解析结果。图4是表示解析结果的显示例的图。在图4的例子中,以人物的脸为对象,进行年龄及性别的识别,其统计由坐标图表示。在将拍摄部11设于店铺的基础上,通过使用这样的解析结果,能够有助于该店铺的销售活动。As shown in FIG. 3 , a result display device 30 is also connected to the image receiving device 20 . The result display device 30 is, for example, a liquid crystal display panel, and displays images and/or analysis results stored in the database 24 . FIG. 4 is a diagram showing a display example of an analysis result. In the example of FIG. 4 , the age and gender are recognized on the face of a person, and the statistics thereof are represented by a graph. After installing the imaging unit 11 in a store, by using such an analysis result, it is possible to contribute to the sales activities of the store.
返回图3,模式设定部25设定由对象解析部23进行的识别认证处理的模式。对象解析部23按照由模式设定部25设定的模式而执行识别认证处理。发送部26将表示由模式设定部25设定的模式的信息(模式信息)向图像发送装置10发送。Returning to FIG. 3 , the mode setting unit 25 sets the mode of the identification and authentication process performed by the object analyzing unit 23 . The object analysis unit 23 executes identification and authentication processing according to the mode set by the mode setting unit 25 . The transmitting unit 26 transmits information (mode information) indicating the mode set by the mode setting unit 25 to the image transmitting device 10 .
图5是说明从图像接收装置20对图像发送装置10的模式的指定的图。如图5所示,从图像接收装置20向图像发送装置10发送模式信息。发送的该模式信息如上述那样,由图像发送装置10的接收部18接收,在削减控制部17的控制中使用。在数据削减部15中,通过与该模式信息对应的削减方法,生成数据量削减图像。如此生成的数据量削减图像从图像发送装置10向图像接收装置20发送。即,图像接收装置20将由模式设定部25设定的识别认证处理的模式信息对于图像发送装置10发送时,从图像发送装置10按照该模式信息适当地选择而削减了数据量的数据量削减图像被发送。FIG. 5 is a diagram illustrating designation of a mode from the image receiving device 20 to the image transmitting device 10 . As shown in FIG. 5 , mode information is transmitted from the image receiving device 20 to the image transmitting device 10 . The transmitted mode information is received by the receiving unit 18 of the image transmitting device 10 as described above, and is used for the control of the reduction control unit 17 . In the data reduction unit 15, a data amount reduction image is generated by a reduction method corresponding to the pattern information. The data volume-reduced image generated in this way is transmitted from the image transmission device 10 to the image reception device 20 . That is, when the image receiving device 20 transmits the mode information of the identification and authentication process set by the mode setting unit 25 to the image transmitting device 10, the image transmitting device 10 appropriately selects and reduces the amount of data in accordance with the mode information. The image is sent.
接下来,详细说明图像的识别认证处理的模式和从图像发送装置10发送的图像的数据量的削减(最佳拍摄的选择、及图像自身的数据量的削减)的方法的关系。如上述那样,在图像发送装置10中,通过最佳拍摄判定部151从多个图像选择一部分,并通过图像数据量削减部152进行削减该选择图像自身的数据量的处理(修剪、重新设定尺寸、编码),由此能够减少网络NW的负荷及图像接收装置20的处理负荷、数据库的必要容量,但对于同一对象,图像接收装置20接收的图像的个数受限制,由此,图像接收装置20中的识别认证处理的精度可能会下降,而且,在图像接收装置20中接收的图像的数据量被削减,识别认证处理的精度也可能会下降。Next, the relationship between the mode of the image recognition and authentication process and the method of reducing the data size of the image transmitted from the image transmission device 10 (optimum shot selection and reduction of the data size of the image itself) will be described in detail. As described above, in the image transmission device 10, the optimal shot determination unit 151 selects a part of a plurality of images, and the image data size reduction unit 152 performs processing (trimming, resetting, etc.) to reduce the data size of the selected image itself. size, encoding), thereby reducing the load on the network NW, the processing load on the image receiving device 20, and the necessary capacity of the database. However, for the same object, the number of images received by the image receiving device 20 is limited. The accuracy of the identification and authentication process in the device 20 may decrease, and the data amount of the image received by the image receiving device 20 is reduced, and the accuracy of the identification and authentication process may also decrease.
因此,在本实施方式的图像识别认证系统100中,以选择对于图像接收装置20的识别认证处理而言适当的图像的方式设定通过最佳拍摄判定部151选择图像时的基准,由此抑制对于同一对象限制图像接收装置20接收的图像的个数所造成的识别认证处理的精度的下降,并且,以在图像接收装置20的识别认证处理所需的范围内削减数据量的方式设定通过图像数据量削减部152削减图像的数据量时的削减方法,由此抑制由图像接收装置20接收的图像的数据量被削减所造成的识别认证处理的精度的下降。Therefore, in the image recognition and authentication system 100 of the present embodiment, the criteria for selecting an image by the optimum shot determination unit 151 are set so that an image suitable for the recognition and authentication process of the image receiving device 20 is selected, thereby suppressing For the same object, the reduction in the accuracy of the recognition and authentication process due to the number of images received by the image receiving device 20 is limited, and the amount of data is reduced within the range required for the recognition and authentication process of the image receiving device 20. The reduction method when the image data volume reduction unit 152 reduces the data volume of the image thereby suppresses the reduction in the accuracy of the recognition and authentication process due to the reduction of the data volume of the image received by the image receiving device 20 .
图6是表示分辨率(像素数)与识别认证的精度的关系的例子的坐标图。如图6所示,在进行脸的检测(作为脸的识别)作为识别认证处理时,只要是12×12左右的分辨率,就能够以80%左右的精度进行判定。而且,在进行性别判定作为识别认证处理时,能够以60×60左右的分辨率且以80%左右的精度进行判定。在进行年龄判定作为识别认证处理时,为了得到80%左右的精度,需要80×80左右的分辨率。如此,对应于识别认证处理的内容为了得到一定的精度所需的分辨率不同。即,在上述的图像识别认证系统100中,在通过图像发送装置10进行的脸的检测处理和通过图像接收装置20进行的性别判定处理或年龄判定处理中,由于必要的分辨率不同,因此不考虑通过图像接收装置20进行的识别认证处理而通过图像发送装置10选择了图像时,对于图像接收装置20的判定而言产生不适合的情况,发生精度劣化。FIG. 6 is a graph showing an example of the relationship between resolution (number of pixels) and accuracy of identification and authentication. As shown in FIG. 6 , when performing face detection (recognition as face) as recognition and authentication processing, as long as the resolution is about 12×12, it can be determined with an accuracy of about 80%. Furthermore, when gender determination is performed as the identification and authentication process, it can be determined with a resolution of approximately 60×60 and with an accuracy of approximately 80%. When performing age determination as identification and authentication processing, a resolution of approximately 80×80 is required in order to obtain an accuracy of approximately 80%. In this way, the resolution required to obtain a certain degree of accuracy differs depending on the content of the identification and authentication process. That is, in the above-mentioned image recognition authentication system 100, in the face detection process performed by the image transmitting device 10 and the gender determination process or age determination process performed by the image receiving device 20, since the necessary resolutions are different, there is no difference between them. Considering the identification and authentication process performed by the image receiving device 20 , when an image is selected by the image transmitting device 10 , it may not be suitable for the determination of the image receiving device 20 and the accuracy may be deteriorated.
从图6可知,在任一识别认证处理中,在一定的分辨率以上,识别认证的精度的上升的程度变缓,由此,即使为一定以上的分辨率,与作用于网络的负荷或作用于识别认证处理的负荷等增加相比,无法期待识别认证的精度的上升。It can be seen from FIG. 6 that in any identification and authentication process, the degree of improvement of the accuracy of identification and authentication becomes slow when the resolution is above a certain level. Therefore, even if the resolution is above a certain level, it has nothing to do with the load on the network or on the network. Compared with an increase in the load of the identification and authentication processing, etc., the improvement of the accuracy of the identification and authentication cannot be expected.
在图像发送装置10的参数存储部16中存储有规定了识别认证处理的内容(模式)与包含分辨率或高频成分的必要度的削减参数之间的对应关系的表格。图7是表示存储在参数存储部16中的表格的例子的表。在图像接收装置20的识别认证处理的模式中存在年龄判定模式、性别判定模式、人物认证模式、时装流行解析模式、体型判定模式、人种判定模式、车种判定模式、号码判定模式、宠物判定模式、及人缩略图显示模式。The parameter storage unit 16 of the image transmission device 10 stores a table defining a correspondence relationship between the content (mode) of the recognition and authentication process and reduction parameters including resolution and necessity of high-frequency components. FIG. 7 is a table showing an example of a table stored in the parameter storage unit 16 . The identification and authentication processing modes of the image receiving device 20 include an age determination mode, a gender determination mode, a person identification mode, a fashion trend analysis mode, a body shape determination mode, a race determination mode, a car type determination mode, a serial number determination mode, and a pet determination mode. mode, and person thumbnail display mode.
在表格中,关于各模式,规定了“检测目标”、“解析目标”、“带域”、“区域”、“分辨率”、及“图像质量”的参数。在图6中,对于“分辨率”,说明了各模式所需的分辨率,但关于“带域”及“图像质量”,可以说同样。即,为了进行一定的精度的识别认证处理而需要高频成分多的图像,但其程度按照脸的检测、性别判定、年龄判定的顺序升高。并且,当高频成分存在的程度增多时,这以上的话,与高频成分的增多相对的识别认证处理的精度的上升变缓。而且,为了进行一定的精度的识别认证处理,需要图像质量高的图像,但其程度按照脸的检测、性别判定、年龄判定的顺序升高。并且,当图像质量变好某种程度时,这以上的话,与图像质量的变好相对的识别认证处理的精度的上升变缓。In the table, parameters of "detection target", "analysis target", "band", "area", "resolution", and "image quality" are specified for each mode. In FIG. 6 , the resolution required for each mode is described for "resolution", but the same can be said about "bandwidth" and "image quality". That is, an image with many high-frequency components is required to perform identification and authentication processing with a certain accuracy, but the degree of this increases in the order of face detection, gender determination, and age determination. In addition, when the degree of presence of high-frequency components increases, the increase in the accuracy of the identification and authentication processing becomes slower relative to the increase in high-frequency components. Furthermore, in order to perform recognition and authentication processing with a certain accuracy, images with high image quality are required, but the degree of this is increased in the order of face detection, gender determination, and age determination. In addition, when the image quality improves to a certain extent, the increase in the accuracy of the recognition and authentication processing relative to the improvement in image quality becomes slower.
参照图7具体说明的话,例如,在年龄判定模式中,首先“检测目标”当然是人。而且,对象解析部23以人物的脸图像为基础来判定该人物的年龄,因此“解析目标”成为脸图像。而且,为了从脸图像来判定年龄而需要脸的皱纹或各部位的详细的位置等细微的信息,因此模糊、抖动少,需要包含较多的高频成分的脸图像,由此,“带域”成为需要高频成分。Specifically, referring to FIG. 7 , for example, in the age determination mode, the "detection target" is, of course, a person first. Furthermore, since the object analysis unit 23 judges the age of the person based on the face image of the person, the "analysis target" becomes the face image. In addition, in order to determine the age from the face image, fine information such as facial wrinkles and detailed positions of each part are required, so there is less blurring and shaking, and a face image containing many high-frequency components is required. ” becomes the need for high-frequency components.
年龄判定所需的区域仅是脸的区域,头发的部分等的信息不需要,因此“区域”成为脸区域。而且,在年龄判定中,如图6所示,需要80×80左右的分辨率,因此“分辨率”成为高分辨率。而且,同样地,为了从脸图像判定年龄,进行细致的解析,因此需要充分的量的信息,图像质量成为高图像质量。需要说明的是,图像质量对应于编码的编码量,高图像质量意味着编码后的编码量多。The area required for age determination is only the face area, and information such as the hair portion is unnecessary, so the "area" is the face area. Furthermore, in age determination, as shown in FIG. 6 , a resolution of about 80×80 is required, so the “resolution” is high resolution. Also, similarly, in order to determine the age from the face image and perform detailed analysis, a sufficient amount of information is required, and the image quality becomes high image quality. It should be noted that the image quality corresponds to the amount of encoded codes, and high image quality means that the amount of encoded codes after encoding is large.
削减控制部17参照存储在参数存储部16中的表格,由此取得与由接收部18接受到的模式信息对应的各参数,对于数据削减部15等,设定取得的参数。具体而言,削减控制部17将“检测目标”设定在对象检测部,将“解析目标”设定在对象特征解析部153,将“带域”设定在带域成分检测部154及必要带域提取部157,将“区域”设定在必要区域提取部156,将“分辨率”设定在分辨率转换部158,将“图像质量”设定在图像编码部159。The reduction control unit 17 refers to the table stored in the parameter storage unit 16 to obtain parameters corresponding to the pattern information received by the reception unit 18 , and sets the obtained parameters to the data reduction unit 15 and the like. Specifically, the reduction control unit 17 sets the “detection target” in the object detection unit, the “analysis target” in the object feature analysis unit 153, the “band” in the band component detection unit 154 and the necessary The band extraction unit 157 sets “area” in the necessary area extraction unit 156 , sets “resolution” in the resolution conversion unit 158 , and sets “image quality” in the image coding unit 159 .
接下来,参照图8,说明图像发送装置10的动作。图8是图像发送装置10中的图像发送方法的流程图。图8所示的流程图通过在图像发送装置10中安装图像发送程序,按照该图像发送程序来执行。Next, the operation of the image transmission device 10 will be described with reference to FIG. 8 . FIG. 8 is a flowchart of an image transmission method in the image transmission device 10 . The flowchart shown in FIG. 8 is executed in accordance with the image transmission program by installing the image transmission program in the image transmission device 10 .
首先,图像输入部12输入拍摄部11拍摄到的图像(步骤S81)。如上所述,从拍摄部11连续地向图像输入部12输入图像,但以下的流程按照每一个图像(帧)反复执行。当输入1帧的图像时,利用对象检测部13检测对象(步骤S82)。此时,应由对象检测部13检测的对象由削减控制部17设定作为“检测目标”。然后,进行由对象追踪部14检测到的对象的追踪(步骤S83)。First, the image input unit 12 inputs an image captured by the imaging unit 11 (step S81 ). As described above, images are continuously input from the imaging unit 11 to the image input unit 12 , but the following flow is repeatedly executed for each image (frame). When an image of one frame is input, an object is detected by the object detection unit 13 (step S82 ). At this time, the object to be detected by the object detection unit 13 is set by the reduction control unit 17 as a “detection target”. Then, the object detected by the object tracking unit 14 is tracked (step S83 ).
接下来,对象追踪部14判断计数值是否达到阈值(步骤S84)。该计数值是无法检测出对象的连续的帧的个数,该阈值是无法检测出对象的连续的帧的容许数。需要说明的是,该计数值对每个对象保持,以下的流程对每个对象进行,例如在画面内存在两个对象时,对于这2个对象分别进行。在计数值未到达阈值时(在步骤S84中为“否”),判断是否检测到对象(步骤S85)。在检测到对象时(在步骤S85中为“是”),将计数值重置(步骤S86),对象特征解析部153进行对象的特征解析(步骤S87)。此时的特征解析的对象通过削减控制部17设定作为“解析目标”。接着,带域成分检测部154检测带域成分(步骤S88)。Next, the object tracking unit 14 determines whether or not the count value has reached a threshold (step S84 ). The count value is the number of consecutive frames in which an object cannot be detected, and the threshold value is an allowable number of consecutive frames in which an object cannot be detected. It should be noted that this count value is held for each object, and the following flow is performed for each object, for example, when there are two objects in the screen, these two objects are respectively performed. When the count value has not reached the threshold (NO in step S84 ), it is determined whether or not an object has been detected (step S85 ). When an object is detected (YES in step S85 ), the count value is reset (step S86 ), and the object feature analysis unit 153 performs feature analysis of the object (step S87 ). The object of feature analysis at this time is set by the reduction control unit 17 as an “analysis target”. Next, the band component detection unit 154 detects the band component (step S88 ).
图9是表示带域成分检测部154的带域成分检测处理(图8的步骤S88)的流程图。带域成分检测部154首先对检测到图像内的对象(步骤S82)的区域的频率进行解析(步骤S881)。带域成分检测部154接着算出通过削减控制部17设定的“带域”的成分量(步骤S882)。FIG. 9 is a flowchart showing the band component detection process (step S88 in FIG. 8 ) performed by the band component detection unit 154 . The band component detection unit 154 first analyzes the frequency of the region where the object in the image is detected (step S82 ) (step S881 ). Next, the band component detection unit 154 calculates the component amount of “band” set by the reduction control unit 17 (step S882 ).
返回图8,接着,图像选择部155进行最佳拍摄的判定(步骤S89)。具体而言,图像选择部155将作为最佳拍摄而保存在暂时保存缓冲器内的图像与此时处理的图像进行比较,基于对象特征解析部153的特征解析的结果及带域成分检测部154的带域成分检测的结果,若为更好的图像(在步骤S89中为“是”),为了将暂时保存缓冲器的图像更新为此时处理的图像,而进行图像数据削减处理(步骤S90)。在保存的图像为更好的图像时(在步骤S89中为“否”),不更新暂时保存缓冲器而结束该图像(帧)的处理。Returning to FIG. 8 , next, the image selection unit 155 determines the best shot (step S89 ). Specifically, the image selection unit 155 compares the image stored in the temporary storage buffer as the best shot with the image processed at this time, and based on the result of the feature analysis by the object feature analysis unit 153 and the band component detection unit 154 If the result of the detection of the band component is a better image ("Yes" in step S89), in order to update the image in the temporary storage buffer to the image processed at this time, image data reduction processing is performed (step S90 ). If the stored image is a better image ("No" in step S89), the processing of the image (frame) ends without updating the temporary storage buffer.
图10是表示图像数据量削减部152中的图像数据量削减处理(图8的步骤90)的流程图。必要区域提取部156按照由削减控制部17设定的“区域”,提取出图像内的必要的区域(步骤S901)。接下来,必要带域提取部157按照由削减控制部17设定的“带域”,提取出必要的带域(步骤S902)。接着,分辨率转换部158按照由削减控制部17设定的“分辨率”,进行图像的分辨率转换(步骤S903)。接着,图像编码部159按照由削减控制部17设定的“图像质量”,对图像进行编码(步骤S904)。FIG. 10 is a flowchart showing image data amount reduction processing (step 90 in FIG. 8 ) in the image data amount reducing unit 152 . The necessary region extracting unit 156 extracts necessary regions within the image according to the “region” set by the reduction control unit 17 (step S901 ). Next, the necessary band extraction unit 157 extracts necessary bands according to the “band” set by the reduction control unit 17 (step S902 ). Next, the resolution conversion unit 158 converts the resolution of the image according to the “resolution” set by the reduction control unit 17 (step S903 ). Next, the image encoding unit 159 encodes the image according to the “image quality” set by the reduction control unit 17 (step S904 ).
返回图8,图像数据量削减部152将存储在暂时保存缓冲器中的图像改写成通过图像数据量削减处理而削减了数据量的图像(数据量削减图像)(步骤S91),结束该图像(帧)的处理。反复进行这样的处理,当对象不再暂时地出现在图像内时,由于在步骤S85中未检测到对象(在步骤S85中为“否”),因此增加计数值(步骤S92),结束该图像(帧)的处理。Returning to FIG. 8 , the image data amount reduction unit 152 rewrites the image stored in the temporary storage buffer into an image (data amount reduced image) whose data amount has been reduced by the image data amount reduction process (step S91 ), and ends the image ( frame) processing. This process is repeated, and when the object no longer temporarily appears in the image, since no object is detected in step S85 ("No" in step S85), the count value is increased (step S92), and the image is ended (frame) processing.
当未检测对象的帧发生连续时,经由步骤S85至步骤S92而结束处理的处理发生连续,如此在未重置计数值的状态下到达规定的阈值时(在步骤S84中为“否”),判断为追踪完成(追踪对象从摄像机之前完全消失),发送部19发送此时保存在暂时保存缓冲器内的图像(数据量削减图像)(步骤S93)。When the frames of the undetected object continue, the process of ending the processing via steps S85 to S92 occurs consecutively, and when the count value has not been reset and reaches a predetermined threshold ("No" in step S84), When it is judged that the tracking is completed (the tracking target has completely disappeared from the front of the camera), the transmission unit 19 transmits the image (data size reduced image) currently stored in the temporary storage buffer (step S93 ).
如以上那样,根据本实施方式的图像识别认证系统100,图像发送装置10的最佳拍摄判定部151选择输入的图像中的最佳拍摄,因此从图像发送装置10向图像接收装置20发送的图像的个数受到限制,由此,能够削减网络的负荷、图像接收装置20的识别认证处理的负荷、图像保存的必要容量。而且,图像发送装置10的图像数据量削减部152对于发送的图像,利用修剪、重新设定尺寸、编码等方法进行削减数据量的处理,因此在这一点上,也能够削减网络的负荷、图像接收装置20内的识别认证处理的负荷、图像保存的必要容量。需要说明的是,也可以仅进行上述的最佳拍摄判定部151的最佳拍摄的选择及图像数据量削减部152的图像数据量的削减中的任一个。As described above, according to the image recognition authentication system 100 of this embodiment, the best shot determination unit 151 of the image transmission device 10 selects the best shot among the input images, so the image transmitted from the image transmission device 10 to the image reception device 20 The number of objects is limited, thereby reducing the load on the network, the load on the identification and authentication processing of the image receiving device 20, and the required capacity for image storage. Furthermore, the image data volume reduction unit 152 of the image transmission device 10 performs processing for reducing the data volume of the transmitted image by methods such as trimming, resizing, and encoding. The load of identification and authentication processing in the receiving device 20 and the necessary capacity for image storage. It should be noted that only one of the selection of the best shot by the best shot determination unit 151 and the reduction of the image data amount by the image data amount reduction unit 152 described above may be performed.
在本实施方式的图像识别认证系统100中,还通过与图像接收装置20中的图像的识别认证处理适合的方法来进行图像发送装置10中的上述的最佳拍摄判定及图像数据量削减的处理,因此不仅能够削减发送的数据的量,而且能够抑制这样的数据量的削减引起的图像接收装置20的识别认证处理的精度的下降。In the image recognition and authentication system 100 of the present embodiment, the above-mentioned optimum shot determination and image data amount reduction processing in the image transmission device 10 are also performed by a method suitable for the image recognition and authentication process in the image reception device 20 Therefore, not only the amount of data to be transmitted can be reduced, but also the reduction in the accuracy of the identification and authentication process of the image receiving device 20 caused by such a reduction in the amount of data can be suppressed.
对于上述的图像识别认证系统100,能够进行各种变形。以下,说明几个其具体例。最佳拍摄判定部151及图像数据量削减部152中的任一者可以省略的情况如上所述。而且,可以仅省略图像数据量削减部152的一部分的结构。Various modifications can be made to the image recognition authentication system 100 described above. Hereinafter, some specific examples thereof will be described. Either one of the optimum shot determination unit 151 and the image data amount reduction unit 152 may be omitted as described above. Furthermore, only a part of the configuration of the image data amount reduction unit 152 may be omitted.
另外,在上述的实施方式中,图像发送装置10具备存储有规定了图像的识别认证的模式和与应发送的数据量的削减方法相关的各种参数之间的对应关系的表格的参数存储部16,通过接收部18从图像接收装置20接收模式信息,而削减控制部17设定了与该模式信息对应的各种参数,但本发明并不局限于此。例如,模式信息可以从图像接收装置20以外接收。而且,可以取代接收部18而具备输入部,模式信息在图像发送装置10中由使用者输入。In addition, in the above-described embodiment, the image transmission device 10 has a parameter storage unit that stores a table that defines the correspondence relationship between the pattern of image recognition and authentication and various parameters related to the method of reducing the amount of data to be transmitted. 16. The mode information is received from the image receiving device 20 through the receiving unit 18, and the reduction control unit 17 sets various parameters corresponding to the mode information, but the present invention is not limited thereto. For example, mode information may be received from outside the image receiving device 20 . Furthermore, an input unit may be provided instead of the reception unit 18 , and the mode information is input by the user in the image transmission device 10 .
另外,在上述的实施方式中,图像选择部155按照由削减控制部17设定的条件仅选择了一个最佳的图像(最佳拍摄),但图像选择部155可以选择最佳的规定个数的图像,也可以选择满足规定的条件的全部图像(个数没有限制)。In addition, in the above-mentioned embodiment, the image selection unit 155 selects only one optimal image (best shot) according to the condition set by the reduction control unit 17, but the image selection unit 155 may select a predetermined number of optimal images. images, you can also select all images that meet the specified conditions (the number is not limited).
而且,图像接收装置20可以具备存储有上述的表格的参数存储部,并将与图像的识别认证处理的模式对应的各种参数向图像发送装置10发送。这种情况下,在图像发送装置10不需要参数存储部16,接收部18接收各种参数,而削减控制部17对其进行设定。图11是表示其变形例的图。Furthermore, the image receiving device 20 may include a parameter storage unit storing the above-mentioned table, and transmit various parameters corresponding to the pattern of the image recognition and authentication process to the image transmitting device 10 . In this case, the image transmission device 10 does not require the parameter storage unit 16 , and the reception unit 18 receives various parameters, and the reduction control unit 17 sets them. FIG. 11 is a diagram showing a modified example thereof.
图像接收装置20当设定模式时,参照表格,读出与该模式对应的各种参数,作为表示削减方法的信息而将这些参数向图像发送装置10发送。在图像发送装置10中,由接收部18接收该信息,削减控制部17按照该信息来设定各种参数,由此,以对应于该参数的削减方法来生成数据量削减图像而向图像接收装置20发送。在图像接收装置20中,对于从图像发送装置10发送的数据量削减图像执行识别认证处理,但此时,假设若识别认证的精度不好,则再将提高识别认证的精度的参数向图像发送装置10指定。图像发送装置10按照由图像接收装置20指定的参数,生成数据量削减图像而向图像接收装置20发送。When setting a mode, the image receiving device 20 refers to a table, reads out various parameters corresponding to the mode, and transmits these parameters to the image transmitting device 10 as information indicating a reduction method. In the image transmission device 10, the receiving unit 18 receives this information, and the reduction control unit 17 sets various parameters according to the information, thereby generating a data amount reduction image with a reduction method corresponding to the parameter and receiving the image to the image. Device 20 sends. In the image receiving device 20, the recognition and authentication process is performed on the data-reduced image sent from the image sending device 10, but at this time, if the recognition and authentication accuracy is not good, then the parameters for improving the recognition and authentication accuracy are sent to the image. Device 10 specified. The image transmission device 10 generates a data volume-reduced image according to parameters specified by the image reception device 20 and transmits the image to the image reception device 20 .
另外,上述的实施方式的图像识别认证系统10经由网络NW将1个图像发送装置10与1个图像接收装置20连接,但系统结构并不局限于此。图12是表示系统结构的另一例的图。在该例子中,图像发送装置10与图像接收装置20不经由网络而直接连接。图13是表示系统结构的又一例的图。如图13所示,多个图像发送装置10与多个图像接收装置20也可以经由网络NW连接。需要说明的是,也可以是仅图像发送装置10为多个,并经由网络NW与1个图像接收装置20连接。In addition, the image recognition authentication system 10 of the above-mentioned embodiment connects one image transmission device 10 and one image reception device 20 via the network NW, but the system configuration is not limited to this. FIG. 12 is a diagram showing another example of the system configuration. In this example, the image transmitting device 10 and the image receiving device 20 are directly connected without going through a network. FIG. 13 is a diagram showing still another example of the system configuration. As shown in FIG. 13 , a plurality of image transmission devices 10 and a plurality of image reception devices 20 may be connected via a network NW. It should be noted that only the image transmission device 10 may be provided in plurality and be connected to one image reception device 20 via the network NW.
以上说明了当前时刻考虑到的本发明的优选实施方式,但对于本实施方式能够进行多样的变形,并且,权利要求书包括存在于本发明的真实的主旨和范围内的这样的全部的变形。The preferred embodiments of the present invention considered at the present time have been described above, but various modifications can be made to the present embodiments, and the appended claims include all such modifications within the true spirit and scope of the present invention.
【工业上的可利用性】【Industrial availability】
本发明具有能够减少网络负荷及服务器的成本,并抑制这样的数据量的削减造成的识别认证处理的精度的下降这样的效果,作为连续输入的图像的识别或进行识别的图像识别认证系统等有用。The present invention has the effect of reducing network load and server cost, and suppressing the reduction in the accuracy of recognition and authentication processing due to such a reduction in the amount of data, and is useful as an image recognition and authentication system for recognizing continuously input images or performing recognition. .
Claims (11)
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2012-160453 | 2012-07-19 | ||
| JP2012160453A JP6024952B2 (en) | 2012-07-19 | 2012-07-19 | Image transmission apparatus, image transmission method, image transmission program, and image recognition authentication system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103577802A CN103577802A (en) | 2014-02-12 |
| CN103577802B true CN103577802B (en) | 2018-08-31 |
Family
ID=
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6680745B2 (en) * | 2000-11-10 | 2004-01-20 | Perceptive Network Technologies, Inc. | Videoconferencing method with tracking of face and dynamic bandwidth allocation |
| CN101990757A (en) * | 2008-04-08 | 2011-03-23 | 富士胶片株式会社 | Image processing system, image processing method, and medium for storing program |
| US8675065B2 (en) * | 2008-12-04 | 2014-03-18 | Hitachi, Ltd. | Video monitoring system |
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6680745B2 (en) * | 2000-11-10 | 2004-01-20 | Perceptive Network Technologies, Inc. | Videoconferencing method with tracking of face and dynamic bandwidth allocation |
| CN101990757A (en) * | 2008-04-08 | 2011-03-23 | 富士胶片株式会社 | Image processing system, image processing method, and medium for storing program |
| US8675065B2 (en) * | 2008-12-04 | 2014-03-18 | Hitachi, Ltd. | Video monitoring system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6024952B2 (en) | Image transmission apparatus, image transmission method, image transmission program, and image recognition authentication system | |
| Agustsson et al. | Generative adversarial networks for extreme learned image compression | |
| US8938092B2 (en) | Image processing system, image capture apparatus, image processing apparatus, control method therefor, and program | |
| US8675065B2 (en) | Video monitoring system | |
| JP5366855B2 (en) | Image processing method and apparatus, and program | |
| JP5335713B2 (en) | Image processing method and apparatus, and program | |
| US10771788B2 (en) | Information processing device and information processing method | |
| KR20080058171A (en) | Camera tampering detection | |
| KR101704775B1 (en) | Apparatus and method for multi-resolution image processing | |
| US20100250588A1 (en) | Image searching system and image searching method | |
| JP2012209831A (en) | Image data control device | |
| CN106060544B (en) | Image coding method, related equipment and system | |
| EP4195166A1 (en) | Apparatus and method for transmitting images and apparatus and method for receiving images | |
| US20130208984A1 (en) | Content scene determination device | |
| TW201328359A (en) | Moving object detection method and apparatus based on compressed domain | |
| CN110249366A (en) | Image feature amount output device, pattern recognition device, image feature amount output program and image recognition program | |
| JP7255841B2 (en) | Information processing device, information processing system, control method, and program | |
| US20250119511A1 (en) | Apparatus and method for transmitting images and apparatus and method for receiving images | |
| JP6784032B2 (en) | Encoding device, coding method, program, and image processing system | |
| CN103577802B (en) | Picture transmitter device, method and program, image recognition Verification System and image received device | |
| JP2019169843A (en) | Video recording device, video recording method and program | |
| JP6720743B2 (en) | Media quality determination device, media quality determination method, and computer program for media quality determination | |
| CN113269014B (en) | Image processing method, device and storage medium | |
| KR101323886B1 (en) | Distributed-Processing-Based Object Tracking Apparatus and Method | |
| Rathgeb et al. | Evaluating the impact of iris image compression on segmentation and recognition accuracy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200120 Address after: Fukuoka, Japan Patentee after: Panasonic i-pro sensor solutions Co.,Ltd. Address before: Osaka Patentee before: Matsushita Electric Industrial Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |
Address after: Fukuoka Patentee after: Panasonic i-PRO sensor solutions Co.,Ltd. Address before: Tokyo, Japan Patentee before: Panasonic i-PRO sensor solutions Co.,Ltd. |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20210122 Address after: Tokyo, Japan Patentee after: Panasonic i-PRO sensor solutions Co.,Ltd. Address before: Fukuoka, Japan Patentee before: Panasonic i-PRO sensor solutions Co.,Ltd. |
|
| CP03 | Change of name, title or address |
Address after: Tokyo Patentee after: Epro (Japan) Co.,Ltd. Address before: Fukuoka Patentee before: Panasonic i-pro sensor solutions Co.,Ltd. |