JPH1132323A - Monitoring system by still image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a moving vector in a picture by using a photographed still image, and to select and record only a necessary picture by providing a means for selecting only a necessary picture including the picture of a moving object, and recording only the necessary picture including the picture of the moving object. SOLUTION: A still image pickup part 12 such as a digital still camera photographs a picture in a prescribed range in each prescribed small time, and successively transmits it to an image analyzing means 17. The image analyzing means 17 analyzes a picture photographed by the still image pickup part 12, and detects the movement of an object to be photographed (moving object) such as an intrude for discriminating the presence or absence of the object to be photographed. The movement of the object to be photographed can be discriminated by detecting the moving vector of the picture. A recording means 18 records a picture photographed by the still image photographing part 12 based on the recording instruction of the image analyzing means 17. Thus, the presence or absence of the object to be photographed such as an intruder in a picture can be monitored, and only a picture including the object to be photographed can be selected and recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a surveillance system using a still image photographing means such as a digital still camera, and more particularly to a surveillance system including a moving object image among images photographed by the still image photographing means. The present invention relates to a monitoring system capable of recording only necessary images.

[0002]

2. Description of the Related Art A predetermined monitoring range is set at an entrance or a passage of a building, a garage, a cash dispenser corner of a bank, or the like.
2. Description of the Related Art A monitoring system for security, management, and the like is known. Generally, such a monitoring system captures a predetermined monitoring range continuously or at predetermined time intervals using a video camera and outputs it to a monitor for monitoring by an observer, or an image captured as necessary. All or images at predetermined time intervals are recorded on a recording medium such as a tape.

[0003]

In a conventional surveillance system for recording a captured video on a recording medium, the video is automatically written irrespective of the presence or absence of a capturing target such as an intruder. Even images that do not show are recorded. As a result, a large-capacity recording medium is required, and frequent replacement of the recording medium is required. Furthermore, in order to check whether or not an image of a subject such as an intruder is captured in the captured video, it is necessary to play back all the video including the video not including the captured target, which takes a lot of time. Was needed. For this reason, the checking operation may take a long time, and even if the object to be photographed is overlooked.

[0004] It is an object of the present invention to provide a surveillance system that detects a motion vector in an image using a still image captured by a digital still camera or the like, and selects and records only a necessary image.

[0005]

In order to solve the above-mentioned problems, a monitoring system (10) of the present invention comprises: a still image capturing section (12) for capturing a still image in a predetermined range at predetermined time intervals; An image analysis unit (17) that analyzes an image captured by the imaging unit (12) and selects only necessary images including an image of a moving object, and an image analysis unit (17) that is selected by the Recording means (18) for recording only necessary images including images is provided.

As a still image pickup unit (12), a CCD,
A MOS type image pickup device can be mentioned, and a digital still camera can be mentioned as a photographing means having this kind of still image pickup section (12). The still image pickup unit (12)
An image in a predetermined range is photographed every predetermined short time Δt (Δt << t), and is sequentially transmitted to the image analysis means (17).

The image analysis means (17) includes a still image pickup section (1).
In step 2), the presence / absence of the photographing target is determined by analyzing the image photographed every time Δt and detecting the movement of the photographing target (moving object) such as an intruder. The motion of the shooting target can be determined by detecting the motion vector of the image. Image analysis means (17) for detecting a motion vector
Divides the image captured by the still image capturing unit (12) into a plurality of detection areas (36), and sets the detection area (3
6) The motion of the representative point (40) during the time Δt is converted into a motion vector based on the image data of at least one representative point (40) in the detection area (36) and the image data at the time t + Δt in the detection area (36). Motion vector detecting means (16), motion vector
The number of detection areas where (MV) is not zero is counted, and it is determined whether or not the number N of counted detection areas is larger than a predetermined number N-TH, and the number of detection areas where the motion vector MV is not zero is determined. A vector processing means (46) for transmitting a recording instruction for at least one image after time t to the recording means (18) only when it is determined that N is larger than the predetermined number N-TH. .

The recording means (18) is for recording an image photographed by the still image pickup section (12) based on a recording command of the image analysis means (17), and is used as a recording medium (52). ,
Examples include a flash memory, a RAM, a mini disk, an optical disk, and a video tape.

[0009]

According to the monitoring system (10) of the present invention,
For example, by observing a motion vector in a photographed image, the presence or absence of a photographing target such as an intruder in the image is monitored, and an image without a photographing target is not recorded,
Since only the image of the object to be photographed can be selected and recorded, unnecessary images are not recorded. Unnecessary images are not recorded, so that the capacity of the recording medium (52) can be reduced, and the operation of confirming the recorded images can be performed quickly and reliably because there are no unnecessary images. Can be performed. Further, since unnecessary images are not recorded, the power consumption required for operating the monitoring system (10) can be reduced as a whole. Therefore, it is possible to operate for a long time with a battery or a small battery without connecting to a household power supply or the like. Therefore, the power supply wiring is not required for mounting the monitoring system (10), and the monitoring system (10) can be easily installed.

[0010]

Embodiments of the present invention will be described below. As shown in FIG. 1, a monitoring system (10) of the present invention digitizes an image captured by a still image capturing unit (12) that captures an image at every minute time Δt, and an image captured by the image capturing unit (12). An A / D converter (14) extracts specific image data (for example, a luminance signal Y) from the digitized image signal, detects a change in the image during Δt, and transmits a necessary image recording command. Image analysis means (17), and a recording means (18) for compressing and recording an image taken by the imaging unit (12) based on a recording command of the image analysis means (17). .

Hereinafter, each of the above components will be described in detail. The imaging unit (12) includes a CCD or MOS type imaging device, and is arranged so as to be able to capture a predetermined monitoring range. The imaging unit captures an image at every minute time Δt and converts the captured image signal into an AD signal.
Send to converter. The minute time Δt is desirably set to the scanning time of one field or one frame or an integral multiple thereof, and can be set to 1/60 seconds, for example.

Image analysis means The image analysis means (17) includes a filter (20) as shown in FIG.
(22), representative point memory (24), correlation calculation unit (26), correlation accumulation unit
(28), a motion vector detecting means (16) comprising an average value accumulating section (30), a minimum value position detecting section (32) and a timing generating section (34)
And a vector processing means (46) for analyzing the detection result of the motion vector detection means (16). Input luminance signal Y
First, the signal passes through a vertical low-pass filter (VLPF) (20) and a horizontal band-pass filter (HBPF) (22) of the motion vector detecting means (16) to remove components other than predetermined frequency components. The timing generator (34) is regulated by a predetermined clock (35), and divides the screen into m × n detection areas (36) (where m and n are both positive integers) as shown in FIG. Then, the luminance signal Y is scanned.

Each detection area (36) is further divided into a plurality of detection sections. In the following description, as shown in FIG. 3, the detection areas are divided into four detection sections (S1 to S1).
An example of division into 4) will be described. Each detection section
(S1 to S4) are composed of s × t pixels (38), and one or a plurality of arbitrary pixels in the section (in this embodiment, pixels at the center of the detection section) are set in advance at the representative point (401). )
(402), (403), and (404) are set. The pixel at the center of the detection area (36) is the representative point (40) of the detection area (36).
Is set as In this embodiment, each detection area (36) is further divided into a plurality of detection sections (S1 to S4), and the average value of the motion vectors of the detection sections belonging to each detection area is obtained. This is to improve the reliability of the motion vector for each detection area.

In the screen luminance signal Y at time t,
The representative point memory (24) stores the luminance signals Y of the representative points (401) to (404) of each section as representative point data for each detection area.

Next, the luminance signal Y of the image at time t + Δt is input to the vector detecting means (16). The input luminance signal Y similarly passes through the filters (20) and (22), removes components other than predetermined frequency components, and is similarly controlled by the timing generator (34) to scan the screen.

The correlation calculator (26) calculates the luminance signal Y for the pixel of each detection section (S1 to S4) at time t + Δt and the stored representative point data at time t in the detection section (S1 to S4). Then, the difference is calculated for each pixel of each detection section, converted into an absolute value, and transmitted to the correlation accumulation unit (28). The correlation accumulation unit (28)
An accumulation memory (42) corresponding to (36) is provided. Each accumulation memory (42) has the same number of pixels as s × the number of pixels in one detection section.
t storage unit (44), and the absolute value of the difference in luminance between the luminance signal Y and the representative point data for each pixel in the detection section is sequentially accumulated and stored in each storage unit (44). , The difference cumulative addition value. In the case of the present embodiment, 4
A value obtained by adding the absolute value of the difference in luminance between the two detection sections is stored.

Next, with respect to the difference cumulative addition value stored in each of the cumulative memories (44), a point having the smallest cumulative value is searched by a minimum value position detecting section (32). Find the difference in coordinates to a point. As shown in FIG. 4, when the accumulated difference values are converted into a three-dimensional graph, the representative points in the detection area are not changed as long as the luminance of each part of the image of the object on the screen does not change during the time Δt. If there is a motion in the image of point V, the coordinate of the point V at which the difference cumulative addition value is the minimum is the representative point
The coordinates are different from the coordinates of (40). On the other hand, when there is no motion at the representative point (40), the point V at which the difference cumulative addition value is minimum is obtained.
Corresponds to the coordinates of the representative point (40). Minimum value position detector (32)
Calculates the difference (sv, tv) between the coordinates of the point V and the representative point A at which the obtained difference cumulative addition value becomes the minimum, and transmits the difference to the vector processing means (46) (step 1 in FIG. 5). Note that the difference (sv, tv) between the coordinates becomes the motion vector of the representative point (40) of the detection area (36).

As will be described later, in order to enhance the reliability of the obtained motion vector, the difference cumulative addition value of each accumulation memory (42) calculated by the correlation accumulation unit (28) is added to the average value integration unit ( 30), and multiply them by the number of pixels (s × t)
To calculate the average difference cumulative addition value (AVE) in the detection area (36). The calculated average difference cumulative addition value (AVE) is similarly transmitted to the vector processing means (46).
(Step 1 in FIG. 5).

FIG. 5 is a flowchart showing the flow of each processing in the vector processing means (46). First, the correlation accumulator (2
The average difference cumulative addition value (AVE) calculated in 8) is compared with a preset average difference cumulative addition threshold value (AVE-TH), and the reliability of the average difference cumulative addition value (AVE) of each detection area is compared. Check sex (step 2). When it is difficult to determine that the representative point has moved to the position where the representative point has actually been detected in the captured image of the detection area with little difference in luminance, the average difference cumulative addition value (AVE) is set to the threshold ( AVE-TH). Specifically, when the detection area corresponds to a portion where a plain wall is photographed, there is almost no difference in luminance between pixels in the detection area. At this time, since it is difficult to accurately determine the movement of the representative point, it is excluded from the following motion vector determination.

In the above steps, the average difference cumulative addition value (A
VE) is determined only for the motion vector of the detection area determined to be smaller than the threshold (AVE-TH).
(Step 3). First, the minimum value among all motion vectors
(MIN) is obtained, and it is confirmed that the minimum value (MIN) is smaller than the minimum value threshold value (MIN-TH). This is for improving the reliability of the obtained motion vector. When the minimum value (MIN) of the motion vector is equal to or more than the threshold value (MIN-TH), the digital still camera shakes for some reason. This is because the captured image may move as a whole or move as a whole, and lack reliability.

In the above step, when the minimum value (MIN) of the motion vector is larger than the minimum threshold value (MIN-TH), it is determined that the reliability of the calculated motion vector is low, and the determination is stopped ( Step 4) A motion vector is calculated for the image input to the motion vector detecting means again. On the other hand, when the minimum value (MIN) of the motion vector is smaller than the minimum threshold value (MIN), it is determined that the reliability of the calculated motion vector is high, and the following calculation is performed.

For the motion vector of the detection area, for each component (sv, tv) of the motion vector MV, the component sv
Alternatively, a motion vector in which at least one of the absolute values | sv | and | tv | of tv is larger than predetermined threshold values sv-TH and tv-TH is detected. When any of the components is larger than the predetermined threshold value, it means that a moving target such as an intruder has been captured in the detection area, and when the component is less than the threshold value, the image of the detection area remains for the time Δt. Does not change, that is, it means that a photographing target such as an intruder is not photographed.

For each detection area, the magnitude of the component of the motion vector is compared with a threshold, and the number N of detection areas having a motion vector larger than the threshold is counted (step 5). When the counted number N of the detected areas of the large motion vector is larger than the predetermined number N-TH, it is determined that the shooting target is moving in the image (steps 6 and 7, hereinafter referred to as “moving image mode”). ), Conversely, a predetermined number N
If it is equal to or less than -TH, it is determined that there is no moving object in the image (steps 6 and 8; hereinafter, referred to as "still mode"). When it is determined that the mode is the "moving image mode" (step 7), the recording means (18) described later
(Step 9). When it is determined that the mode is the still mode, the motion vector is obtained for the images at the time t + Δt and the time t + 2Δt in the same manner as described above without recording the image, and the moving image mode and the still mode are determined.

Recording means Recording means (18), as shown in FIG. 1, controls an image compression means (50) for compressing an image by a known method such as JPEG and a signal writing to a recording medium (52). It can be composed of a writing control means (54) and a recording medium (52) such as a flash memory, a RAM, and a mini disk. Recording means
(18) is an image captured by the imaging unit (12) at time t + 2Δt upon receiving a write command from the vector detection unit (16) when the vector detection unit (16) determines that the moving image mode is set. Is compressed and recorded on the recording medium (52). A time stamp may be added to the recorded image as needed. The image to be recorded is the image at time t + 2Δt,
This is because a memory for recording the image at the time t or the time t + Δt is not required. If Δt is set to a small amount of time change of about 1/60 second, the object to be captured is lost even if the recording is not performed at the same time. This is because there is no significant change in the image.

When it is determined that the moving image mode is set and the image is recorded on the recording medium (52), the next motion vector is detected after a lapse of several seconds to about ten seconds (wait time) after the recording. It is desirable to do this afterwards. This is because when a captured image is later checked, even if an image for each Δt is captured, a change in the movement of a capturing target such as an intruder during Δt cannot be visually determined. This is because almost the same image is taken. Also, by increasing the photographing interval, the capacity of the recording medium (52) can be saved.

The recorded image can be reproduced by a known reproducing means.
(60), for example, as shown in FIG. 1, a read control means (62) for reading a recording medium (52), an image demodulation means (64), and a DA converter (66). It can be implemented by outputting to a monitor (not shown) or the like via an external device. The regenerating means (60) may be provided integrally with the monitoring system (10), or may be provided separately.

[0027]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A monitoring system (10) having the above configuration is used for
An embodiment applied to the monitoring of the entry and exit of a person (70) will be described. FIG. 6 shows the inside of the room (70) of the building.
A door (74) is formed on one wall (72) of the room (70),
Inside the room (70) is a desk and chair (76). Note that the room (70) has a time t + Δt from the time t (FIG. 7B).
(FIG. 7C), as the moving object from the door, "Intruder (8
0) ”. 7 and the following description, it is assumed that an image is divided into 4 × 4 detection areas (36) to obtain a motion vector. If the motion vector is zero, the motion vector MV is represented by a point. If not, the motion vector MV is represented by an arrow corresponding to the motion direction. In the following, the “moving image mode” when a non-zero motion vector is detected in two or more detection areas, and when the detection area where a non-zero motion vector MV is detected is 1 or 0 , "Still mode".

A monitoring system (10) including a still image pickup unit (12) is mounted on a ceiling (78) of a room (70).
The door (74) and the area around the door can be photographed. The imaging range (monitoring range) of the still image imaging unit (12) is as shown in FIG. 7A.

Between the images at time t-Δt (FIG. 7A) and time t (FIG. 7B), there is no moving object such as the intruder (80) inside the room, and the desk, chair (76) Since only a stationary object such as a moving object is photographed, the motion vector MV of each detection area is zero as shown in FIG. Is determined. Therefore,
No image is recorded.

From time t (FIG. 7B) to time t + Δt (FIG. 7C)
, Open the door (74) and the intruder (80) enters,
As shown in FIG. 7 (C-1), a non-zero motion vector MV is detected in a detection area (region α in the figure) corresponding to the door portion. Since the number of detection areas in which a motion vector having a size other than zero is detected is four, the image analysis means (17) determines that the moving image mode is set, and the time t + Δ
The 2t image (FIG. 7D) is recorded on the recording medium.

After recording the image, the motion vector MV is detected (FIGS. 7E, F and F-1) in the same manner as described above with a wait time of several seconds, and the non-zero motion vector MV is obtained. If two or more detection areas are detected, the image
The recording shown in FIG. 7G is performed.

Other Embodiments The same control as described above can be performed by attaching the motion vector detecting means (16) of the present invention to a commercially available digital still camera (90). In this case, as shown in FIG. 8, the external video output terminal of the digital still camera (90)
(92), an image analysis means (17) may be connected, and when a motion vector is detected, the external input shutter (94) of the digital still camera (90) may be turned on to capture an image. . The input of the image to the image analysis means (17), the image input from the imaging unit (12) to the AD converter (14), without passing through the recording means (18), the selection circuit (96) and What is necessary is just to output directly to the image analysis means (17) via the DA converter (66). When playing back recorded images, use the external video output terminal (9
2) connect a monitor (not shown) or the like, read control means (62), read and demodulate the image recorded by the image demodulation means (64), a selection circuit (96) and a DA converter (66) And output via. When a digital still camera having a built-in liquid crystal monitor or the like is used, the recorded image may be reproduced on the built-in monitor.

Further, when a digital still camera is placed on a turntable (not shown) or the like, and an object such as an intruder is recognized as a motion vector, a motion vector in a detection area where the motion vector is not zero is detected. The photographing target may be tracked by rotating the turntable in a direction approaching zero.

When the present invention is applied to a digital still camera including a MOS type image pickup device as the image pickup section (12), an area having a motion vector having a size other than zero is used.
It is also possible to take out and record only the image (the area α surrounded by the dotted line in FIG. 7D).

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a monitoring system according to the present invention.

FIG. 2 is a block diagram of an image analysis unit.

FIG. 3 is a conceptual diagram in which an image is divided by a detection area, and the detection area is further divided into detection sections.

FIG. 4 is a three-dimensional graph of a cumulative difference value.

FIG. 5 is a flowchart showing the flow of processing in a vector processing means.

FIG. 6 is a perspective view showing a mounting form of the monitoring system of the present invention.

FIG. 7 is an explanatory diagram showing an image of the monitoring system, a detected motion vector, and an image to be recorded.

FIG. 8 is a block diagram showing another embodiment of the monitoring system.

[Explanation of symbols]

 (10) Surveillance system (12) Still image capturing unit (16) Motion vector detecting means (17) Image analyzing means (18) Recording means (46) Vector processing means

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Haruhiko Murata 2-5-5 Keihanhondori, Moriguchi-shi, Osaka Sanyo Electric Co., Ltd.

Claims (4)

[Claims] 1. A still image capturing section (12) for capturing a still image in a predetermined range at regular intervals, and an image captured by the still image capturing section (12) is analyzed to include an image of a moving object. Image analysis means (17) for selecting only the necessary images, and recording means (18) for recording only necessary images including the image of the moving object selected by the image analysis means (17). Surveillance system based on still images. 2. An image analysis means (17), comprising:
The still image monitoring system according to claim 1, wherein a motion in the image captured in (12) is detected, and only a necessary image including an image of the moving object is selected. 3. The image analysis means (17) divides an image captured by the still image capturing section (12) into a plurality of detection areas (36), and divides at least one of the detection areas (36) at time t. Based on the image data of two representative points (40) and the image data of the detection area (36) at time t + Δt (where Δt << t), the motion of the representative point (40) during the time Δt is converted into a motion vector. A motion vector detecting means (16) that counts the number of detection areas where the motion vector (MV) is not zero, and counts the number N of detected detection areas to a predetermined number N-TH
It is determined whether or not the number N of the detection areas in which the motion vector MV is not zero is larger than the predetermined number N-TH. Processing means for transmitting a recording instruction for at least one image of
The still image monitoring system according to claim 2, comprising (46). 4. The recording means (18) comprises a vector processing means (46).
, The number N of detection areas where the motion vector MV is not zero
Is determined to be greater than the predetermined number N-TH, the time t
The still image monitoring system according to claim 3, wherein an image at + 2Δt is recorded.