KR20210084946A - Method and server for inspecting food advertising images - Google Patents

Abstract

The food advertisement image inspection method according to an embodiment of the present invention is a method for the image inspection server to inspect food advertisement images, the steps of collecting food advertisement images published on a commerce site, and converting the collected food advertisement images into the original advertisement image and, searching for changes in the collected food advertisement image and determining whether to modify the collected food advertisement image according to the search result.
According to the present invention, by correcting food advertisement images uploaded to a commerce site to the most recently updated food advertisement image, it is possible to provide only accurate advertisement information to consumers and to secure customer trust.

Description

Food advertisement image inspection method and server {METHOD AND SERVER FOR INSPECTING FOOD ADVERTISING IMAGES}

The present invention relates to a food advertisement image inspection method and server, and more particularly, by comparing and analyzing a food advertisement image exposed through a commerce site and accessible to consumers with an original food advertisement image, each commerce site obtains the correct food advertisement image. It relates to a method of checking for publication and a server that performs it.

A food company that operates an F&B (Food & Beverage) market creates a product advertisement image for the products it sells, and distributes it to each commerce site. Among them, when a brand strategy is changed or a product expression regulation is applied, a food advertisement image may cause an over-advertisement problem. Therefore, marketing phrases, ingredients, country of origin, etc. are frequently corrected.

In addition, in recent years, not only have customers' interest in food and health supplement ingredients high, but also accusations are being made on foods that make false and exaggerated advertisements to the Ministry of Food and Drug Safety, Food Country, and the Consumer Agency. Advertisement images should be reviewed periodically.

Accordingly, each food company periodically inspects the information described in the advertisement image and distributes the modified advertisement image to each commerce site. However, the number of products produced and sold by one brand is large, and the number of commerce sites distributing them Since there are more than 100, it takes a lot of time to verify that the seller has posted the correct ad image.

In addition, food advertisement images, unlike product advertisement images such as home appliances and cosmetics, contain a lot of text and product pictures such as marketing phrases, product pictures, cooking examples, recipes, ingredients, etc., so the image size is large and the composition is complicated. There is a limit for the manager to visually inspect it.

In this regard, as a prior art, a method for monitoring an error of an image distributed on a site has been disclosed, but the prior art simply collects an image distributed on a site, and then checks the loading time of the image and whether the image is output As such, it cannot catch errors in the composition and content included in the image.

Therefore, it is required to develop a new technology capable of finding and correcting an incorrect advertisement image in a food advertisement image distributed on a commerce site, and the present invention relates to this.

Korean Patent Publication No. 10-1655800 (2016.09.02.)

The technical problem to be solved by the present invention is to collect food advertisement images exposed to users for food sales on many commerce sites, and compare each advertisement image with the original advertisement image to correct areas that are incorrectly marked. It is to provide a method for inspecting food advertisement images.

Another technical problem to be solved by the present invention is to divide food advertisement images collected from commerce sites into smaller pieces to detect comparison items, determine the types of objects such as colors, borders, and texts in the divided area, and perform specific image comparison. It is to provide a food advertisement image inspection method that can be used.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

The food advertisement image inspection method according to an embodiment of the present invention is a method for the image inspection server to inspect food advertisement images, the steps of collecting food advertisement images published on a commerce site, and combining the collected food advertisement images with the original advertisement image. By comparison, searching for changes in the collected food advertisement image and determining whether to modify the collected food advertisement image according to the search result.

According to an embodiment, the collecting of the food advertisement image includes: inputting at least the name of the food and the original advertisement image among the name of the food that needs to be inspected, the filtering phrase, and the original advertisement image, the name of the food and the filtering phrase Accordingly, the method may further include the steps of searching for a corresponding web page and storing the corresponding food advertisement image in the searched web page. In storing the image, an image excluding unnecessary images may be stored.

According to an embodiment, the step of searching for the change may include extracting a reference point of the collected food advertisement image based on the original advertisement image, and using the extracted reference point, the inspection area within the collected food advertisement image It may further include the step of designating.

According to one embodiment, after the step of designating the inspection area, the step of confirming the comparison type of the inspection area and the step of pre-processing the inspection area according to the confirmed comparison type may be further included.

According to an embodiment, the step of designating the inspection area may further include dividing the original advertisement image and the inspection area into the same plurality of sections.

According to an embodiment, the confirming of the comparison type may include recognizing at least one image classification item among a background color, an image outline, and a character within the original advertisement image or the inspection area, and one object in the recognized image item. It may further include the steps of grouping and determining whether one grouped object is a character, and classifying the comparison type for the inspection area into a character type or a non-character type.

According to an embodiment, the checking of the comparison type may include, if the comparison type is not a text type, extracting a feature point from each of the original advertisement image and the inspection area and the inspection area based on each characteristic point It may further include the step of retrieving the change in .

According to an embodiment, the confirming of the comparison type may further include recognizing a word in each of the original advertisement image and the inspection area and searching for a change when the comparison type is a text type. have.

According to an embodiment, the determining of whether to modify may include generating a modified food advertisement image by displaying the change in the collected food advertisement image when the change is found, and the modified food advertisement image The method may further include transmitting to a commerce server that manages the commerce site.

The image inspection server according to another embodiment of the present invention includes an image collection unit that collects food advertisement images published on a commerce site, compares the collected food advertisement image with an original advertisement image, and changes in the collected food advertisement image. and an image analysis unit that searches for , and an image editing unit that determines whether to modify the collected food advertisement image according to the search result.

According to an embodiment, the image collection unit includes an input unit for receiving at least the name of food and an original advertisement image among the name of the food requiring inspection, the filtering phrase, and the original advertisement image, the name and filtering phrase of the food in the commerce site. Accordingly, it may further include a search unit for searching a corresponding web page and a storage unit for storing a food advertisement image of the corresponding food in the searched web page. Optionally, the storage unit may store images excluding unnecessary images.

According to one embodiment, the image analysis unit, using the feature point extraction unit for extracting the reference point of the collected food advertisement image based on the original advertisement image and the extracted reference point, designate the inspection area in the collected food advertisement image It may further include an image aligning unit.

According to an embodiment, it may further include a type classification unit for confirming the comparison type of the inspection area, and an image preprocessor for pre-processing the inspection area according to the confirmed comparison type.

According to an embodiment, the image alignment unit may divide the original advertisement image and the inspection area into the same plurality of sections.

According to an embodiment, the type classification unit recognizes at least one image classification item of a background color, an image outline, and a character within the original advertisement image or the inspection area, and groups one object in the recognized image classification item , it is possible to determine whether one grouped object is a character, and classify the comparison type for the inspection area into a character type or a non-character type.

According to an embodiment, the feature point extraction unit, when the type classified by the type classification unit is not a text type, extracts a feature point from each of the original advertisement image and the inspection area, and the image analysis unit, based on each feature point to search for changes in the inspection area.

According to an embodiment, the image analysis unit further comprises a text recognition unit for recognizing a word in each of the original advertisement image and the inspection area when the type classified by the type classification unit is a character type, wherein the text recognition unit recognizes You can search for changes in one word. According to an embodiment, when the change is found, the image editing unit generates a modified food advertisement image by displaying the change in the collected food advertisement image, and applies the modified food advertisement image to the commerce site. It may further include a communication unit for transmitting to the commerce server that manages the.

According to the present invention, by correcting food advertisement images uploaded to a commerce site to the most recently updated food advertisement image, it is possible to provide only accurate advertisement information to consumers and to secure customer reliability.

In addition, by dividing the food advertisement image into a plurality of sections, by extracting representative feature points in each divided section, corrections that are difficult to detect can be easily found using the entire image file, and food uploaded to numerous commerce sites Advertising images can be inspected quickly, and food promotion management personnel can be efficiently operated.

Effects of the present invention are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a diagram showing the configuration of an image inspection system according to an embodiment of the present invention.
2 is a diagram showing the configuration of an image inspection server according to an embodiment of the present invention.
3 is a view for explaining a method in which the image inspection server aligns images according to an embodiment of the present invention.
4 is an exemplary view for explaining a method of analyzing an image by the image inspection server according to an embodiment of the present invention.
5 is a diagram illustrating an exemplary modified food advertisement image generated by the image inspection server according to an embodiment of the present invention.
6 is a flowchart illustrating a method for an image inspection server to inspect a food advertisement image according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention, and a method for achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments published below, but may be implemented in various different forms, and only these embodiments make the publication of the present invention complete, and common knowledge in the technical field to which the present invention belongs It is provided to fully inform the possessor of the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used with the meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless clearly defined in particular. The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present invention. As used herein, the singular also includes the plural unless specifically stated otherwise in the phrase.

As used herein, “comprises” and/or “comprising” refers to a referenced component, step and/or action indicating the presence or addition of one or more other components, steps and/or actions. do not exclude

1 is a diagram showing the configuration of an image inspection system 10 according to an embodiment of the present invention.

Referring to FIG. 1 , the image inspection system 10 may include a user terminal 100 , an image inspection server 200 , and a commerce server 300 .

The user terminal 100 may be a terminal possessed by a product manager who wishes to request product advertisement image inspection. The manager uses a web or mobile application capable of executing the image inspection service provided by the image inspection server 200 through the user terminal 100 to manufacture and sell products (hereinafter, the product is described as food) You can check whether the advertisement image of ' has been uploaded correctly to a number of commerce sites.

To this end, the product manager may transmit the name of the food to be inspected or the original advertisement image of the food to the image inspection server 200 through the user terminal 100 . For example, the name of food means the official name of a product such as “Hetbahn Bibimbap” and “Hetbahn Cup Rice”, and the original advertising image of food is an image produced by each company and distributed to each commerce server 300 . it means.

The image inspection server 200 may inspect a food advertisement image published on a commerce site operated by the commerce server 300 . Specifically, the image inspection server 200 may receive a food image inspection request from the user terminal 100, and may perform an image crawler and image analysis. Here, the image crawler is a process of collecting food advertisement images from a commerce site operated by the commerce server 300 based on the name of the food input from the user terminal 100 or the original advertisement image of the food. It collects information that can be defined as "information", can perform functions defined as "image collection, image filter", and image analysis is the process of finding different parts of a collected food advertisement image, "image sorting, type Classification, image text recognition, feature point extraction” can be performed. However, specific functions performed by the image inspection server 200 will be described later.

The commerce server 300 is a server that provides an online sales platform to sellers so that various kinds of food can be sold online, and may manage the sellers or directly sell food. Accordingly, the commerce server 300 may provide a platform through which multiple sellers can sell the same food. For example, the commerce server 300 may manage information about the first to n-th sellers selling company A-a food.

So far, the image inspection system 10 according to an embodiment of the present invention has been described. Hereinafter, the image inspection server 200 that inspects food advertisement images posted on commerce sites by many sellers through the commerce server 300. to be described in detail.

2 is a diagram showing the configuration of the image inspection server 200 according to an embodiment of the present invention.

Referring to FIG. 2 , the image inspection server 200 may include an image collection unit 210 , an image analysis unit 220 , an image editing unit 230 , and a communication unit 240 .

The image collection unit 210 may collect food advertisement images published on commerce sites, and for this purpose, the image collection unit 210 may include an input unit 211 , a search unit 213 , and a storage unit 215 . have. Specifically, the input unit 211 may receive the name or original advertisement image of a food that needs to be inspected from the manager, and may receive an input of a commerce site to be searched. Also, the search unit 215 may search a web page corresponding to the name of the food or the original advertisement image in the commerce site. Here, the search unit 215 may limit the search for unnecessary images through additional phrases and some image filtering processing, and the inspection process for determining whether the corresponding image is the same image is performed by the image analysis unit 220 to be described later can do. In addition, the storage unit 215 may extract and store the food advertisement image of the patdang food from the searched web page. That is, the storage unit 215 may collect content (images in the present invention) existing in a plurality of web pages collected by the search unit 215 .

The image analysis unit 220 compares the collected food advertisement image with the original advertisement image to search for changes in the collected food advertisement image. For this, the image analysis unit 220 includes the feature point extraction unit 221 . , an image alignment unit 223 , a type classification unit 225 , an image preprocessor 227 , and a text recognition unit 229 .

However, the food advertisement image posted by the seller can be arbitrarily edited to fit the image size and capacity according to the characteristics of the commerce site, and new content is added according to the event (eg, delivery information, special price information) conducted by the seller or the commerce site. or can be edited, the image analysis unit 220 may perform a process of aligning the collected food advertisement images for comparison with the original advertisement image.

Specifically, the feature point extraction unit 221 extracts the reference point of the original advertisement image and the reference point of the collected food advertisement image, and the image alignment unit 223 maps each of the extracted reference points to the inspection area within the collected food advertisement image. can be specified.

In this regard, FIG. 3 is a diagram for explaining a method in which the image inspection server 200 aligns images according to an embodiment of the present invention.

Referring to FIG. 3 , the feature point extraction unit 221 of the image inspection server 220 may use a known matching algorithm while adjusting the size of the collected food advertisement image, and finally a reference point ( C1) can be specified. In other words, in the case of images having different sizes, the feature point extraction unit 221 is a correlation coefficient between the reference point (C1) and the reference point (P1) of the original advertisement image for each size of the food advertisement image collected through size conversion of the template image. , and make it possible to compare identical images through a transformation that maximizes the correlation coefficient.

Accordingly, the reference point (C1) for designating the inspection area in the collected food advertisement image may be the four corner portions shown in FIG. 3, and the image alignment unit 223 connects the four reference points (C1) to one You can align the inspection area (A) of In addition, if detailed classification is required, it is possible to compare and align the reference points with one or more images through the division of the inspection area (A).

Again, it will be described with reference to FIG. 2 .

In addition to simply showing the outside/inside of the product, the food advertisement image inspected in the present invention contains various information such as marketing phrases, images of food cooked, recipes, etc., and thus can be generally produced as a vertically long image. Accordingly, the image aligning unit 223 divides the original advertisement image and the inspection area into the same plurality of sections, thereby reducing a comparison error that may occur when comparing a single large image, and improving accuracy. Here, dividing the same plurality of sections may be understood as dividing the original advertisement image and the inspection area into a specified NxN size.

Meanwhile, the type classification unit 225 may group one image connected to each other in the inspection area of the food advertisement image in order to recognize unrecognizable characters in the food advertisement image. For example, the type classification unit 225 may recognize an image outline using a color, group one object included in the recognized image outline, and binarize an image including various colors into black and white, It is also possible to create a boundary group in , and to detect one group in an image through various image processing such as removing unnecessary lines.

That is, the type classification unit 225 first recognizes at least one image classification item among a background color, an image outline, and a character, and secondly determines whether an object (a detected group) is a character or an image in the recognized image item. and, accordingly, the comparison type of the food advertisement image may be classified as a text type or a non-text type. Here, the non-text type may be a color type in which only a background color is displayed, a line type in which an outline/border is displayed, or a standard type.

Also, since the type classification unit 225 separates the background color and the image outline, a separate box-processed content for highlighting in the image may be recognized.

If the type classified by the type classification unit 225 is not a text type, a characteristic point may be extracted from each of the original advertisement image and the inspection area through the characteristic point extraction unit 221 , and the image analysis unit 220 may each Changes can be searched for based on the characteristic points of . At this time. The feature point extraction unit 221 may extract detailed feature points by dividing the inspection area into a plurality of sections.

Conversely, when the type classified by the type classification unit 225 is a character type, a word is recognized in each of the original advertisement image and the inspection area using the text recognition unit 229, and the image analysis unit 220 is the recognized word You can search for changes in .

In addition, the image preprocessor 227 may modify the collected food advertisement image to match the type classified by the type classification unit 225 so that changes can be more easily found. For example, the image pre-processing unit 227 can make the color difference between the background and other images clear by giving a high weight to the central pixel in the color background, and in the case of the outline/border, the image of the thick line is designated as an inspection area The resolution can be increased as much as the converted ratio at the time of writing, and in the case of line characters, only the text area can be clearly seen by deleting the surrounding borders.

In addition, the text recognition unit 229 may utilize Text Detection, Text Recognition, and OCR (Optical Character Recognition) functions to recognize a word, and as the text recognition unit 229 recognizes a word, the image inspection server 200 ) can inspect whether there is a change in the shape and size of the text in the food advertisement image posted by the seller against the intention of the producer.

In addition, the text recognition unit 229 may use a known edit distance and language model to compensate for errors in character recognition depending on the font, character size, and character spacing, and according to the probability value using this, Characters can be recognized accurately.

4 is an exemplary diagram for explaining a method of analyzing an image by the image inspection server 200 according to an embodiment of the present invention.

Referring to FIG. 4 , the food advertisement images collected by the feature point extraction unit 221 and the image alignment unit 223 may be aligned, and the type classification unit 225 shows the type of the aligned food advertisement image. Likewise, it can be distinguished by background color, image outline, and character type.

In addition, the type classification unit 225 may group one object from the divided background color, image outline, and text type, and distinguish images or text from the grouped objects. Accordingly, the feature point extracting unit 221 may extract features from each image or character, and the text recognition unit 229 may select "Purified water, kimchi (made in China), sausage, and red pepper powder??" in the collected food advertisement image. It can recognize characters such as

Referring back to FIG. 3 , the image editing unit 230 may determine whether to correct the collected food advertisement image according to the search result. If the change is found, the image editing unit 230 may generate one modified food advertisement image by displaying the change in the collected food advertisement image.

In this regard, FIG. 5 is a view exemplarily showing a modified food advertisement image generated by the image inspection server 200 according to an embodiment of the present invention.

Referring to FIG. 5 , the image editing unit 230 may provide a modified food advertisement image together with an original advertisement image so that a seller who sells food through a commerce site can recognize at a glance the matters that need to be corrected. Here, reference numbers a1 to a4 may be displayed in the corrected area, and when the seller clicks on it, a detailed description of the change may appear in the form of a pop-up window. For example, a specific description of a change could include something like "The order of highlighting has changed", "The image color has changed", "Delicious > Delicious".

Referring back to FIG. 3 , the communication unit 240 may communicate with the user terminal 100 and the commerce server 300 to exchange changes to the food advertisement image. Specifically, the communication unit 240 may transmit the food advertisement image modified by the image editing unit 230 to the commerce server 300 that manages the commerce site, and the commerce server 300 transmits it to the seller, and then the seller It is possible to transmit the feedback information received from the image inspection server 200 again.

So far, the image inspection server 200 according to an embodiment of the present invention has been described. According to the present invention, the image inspection server 200 collects food advertisement images edited for marketing by each seller and food advertisement images of a previous version, and compares the food advertisement image with the original advertisement image so that the food advertisement image published on the commerce site is the most recent. It can be modified so that it can be a food advertisement image produced in , and accordingly, accurate advertisement information can be provided to consumers. In particular, in the case of a food advertisement image, since various images and characters are mixed, the difference from the original advertisement image can be accurately detected by dividing the image and analyzing it.

6 is a flowchart illustrating a method for the image inspection server 200 to inspect a food advertisement image according to an embodiment of the present invention.

This is only a preferred embodiment for achieving the object of the present invention, and it goes without saying that some steps may be deleted or added, or any one step may be included in another step as needed.

Referring to FIG. 6 , the image inspection server 200 receives a name of food or an original advertisement image from a user, that is, a product manager (S101). In addition, the product manager can specify a commerce site to inspect food advertisement images, such as “CJ MALL” and “HMALL”, and input the commerce site to inspect. In addition, the image inspection server 200 may limit the search for unnecessary images through additional phrases and some image filtering processing, and may receive filtered phrases from the product manager in addition.

After step S101, the image inspection server 200 searches the web page corresponding to the name of the food or the original advertisement image in the web page (S102), edits unnecessary images (S103), and collects only comparable images (S104). Here, the unnecessary image may refer to a representative image of a product disclosed at the top of the web page, rather than an image including a detailed description in the web page.

After step S104, the image inspection server 200 designates the inspection area of the collected food advertisement image (S105). In other words, in addition to images containing descriptions of food, websites that sell food may contain notices from the seller or information about other products sold in bundles by the seller, and the size may be adjusted, so image inspection The server 200 may designate the inspection area of the food advertisement image collected for comparison with the original advertisement image.

Specifically, the image inspection server 200 may measure the correlation coefficient with the original advertisement image while adjusting the size of the collected food advertisement image, and may designate the size at which the correlation coefficient is maximized as the inspection area. That is, the image inspection server 200 adjusts the collected multiple food advertisement images to a size that can be matched with the original advertisement image, so it can be said that the collected food advertisement images are aligned.

After step S105, the image inspection server 200 checks the comparison type of the collected food advertisement image (S106). Here, the comparison type may be largely divided into a character type and a non-character type, and the non-character type may be a color type in which only a background color is displayed, a line type in which an outline/border is displayed, and a standard type.

To check the comparison type in step S106, the image inspection server 200 may group one image connected to each other in the inspection area of the food advertisement image, and determine whether the grouped object is a character (S107).

Specifically, step S107 may be divided into two steps.

Stage 1.

The image inspection server 200 may recognize at least one image classification item of a background color, an image outline, and a character in the inspection area of the food advertisement image, and group one object in the recognized image classification item. In addition, the image inspection server 200 pre-processes the image according to the type after determining the type of the food advertisement image as in step 1 so that changes can be more easily searched for after recognizing any one classification item. can For example, the image inspection server 200 may give a high weight to the central pixel in the color background to make the color difference between the background and other images clear, and in the case of the outline/border, the image of the thick line is designated as the inspection area The resolution can be increased as much as the converted ratio at the time of writing, and in the case of line characters, only the text area can be clearly seen by deleting the surrounding borders. Step 2.

The image inspection server 200 may distinguish whether one grouped object is an image or a text, and accordingly may search for changes in different ways.

As a result of the determination in step S107, if the grouped object is a character, the image inspection server 200 can recognize the character in the food advertisement image by using Text Detection, Text Recognition, and OCR (Optical Character Recognition) functions (S108) , YES), on the contrary, when the grouped object is not a character, by extracting a feature point in the food advertisement image (S109, NO), it is possible to search for changes in the collected food advertisement image.

On the other hand, the image inspection server 200 may divide one food advertisement image into small pieces for a more specific search, and search for changes within the divided area. Specifically, the image inspection server 200 may divide the original advertisement image and the collected food advertisement image into the same plurality of sections. In other words, since the food advertisement image can include long cooking methods in addition to the food image, the original advertisement image and the collected food advertisement image can be divided into NxN sizes smaller than the original size for accurate analysis. You can also search for changes by section.

After steps S108 and S109, the image inspection server 200 searches for changes in whether the collected food advertisement image is the same as the original advertisement image (S110).

When changes are found in step S110, the image inspection server 200 may generate a modified food advertisement image by displaying the changes in the collected food advertisement image (S112, NO), and the category of the corresponding food advertisement image can be classified (S111). Here, the category may be a category defined based on a change such as “check handwriting”, “partly different”, “inconsistent”, “same”, and the like.

Conversely, if the change is not found in step S110, the image inspection server 200 may classify the category of the collected food advertisement image (S111, YES). In this case, the food advertisement image for which the change is not searched may be classified into the "same" category.

After step S111, the image inspection server 200 may determine whether to proceed with the action on the food advertisement image collected based on the classified category (S113).

In step S113, if the category is a category in which changes such as "check handwriting", "partially different", "inconsistent", "same", etc. exist, the image inspection server 200 determines the web page manager on which the corresponding food advertisement image is posted; That is, it is possible to transmit a management notification along with the modified food advertisement image to the commerce server 300 (S114, YES).

Conversely, if the category in step S113 is "matched", the image inspection server 200 sends the image inspection request to the user terminal 100 to the seller of the web page on which the corresponding food advertisement image is posted or the change to the commerce site. does not exist, and only the correct advertisement image is publicly available.

So far, the method for the image inspection server 200 according to an embodiment of the present invention to inspect the food advertisement image has been described. According to the present invention, a company that manages food through a commerce site can use the image inspection server 200 to check whether the food advertisement image distributed and published through a plurality of sellers is the correct food advertisement image, If a food advertisement image exists, it can be corrected to secure customer trust.

In addition, considering that the image is large due to the nature of the advertisement for food, the collected food advertisement image can be divided, and changes can be searched for in each of the divided sections, and characters can be grouped using color within the image. Detect and search for changes within text. That is, by using the image inspection server 200 of the present invention, an administrator can quickly search for changes that are difficult to visually check, so that food promotion management personnel can be operated more efficiently.

Meanwhile, the present invention can also be embodied as computer-readable codes on a computer-readable recording medium. The computer-readable recording medium includes all storage media such as magnetic storage media and optical readable media. Also, it is possible to record the data format of the message used in the present invention on a recording medium.

Although embodiments of the present invention have been described above with reference to the accompanying drawings, those of ordinary skill in the art to which the present invention pertains can realize that the present invention can be implemented in other specific forms without changing the technical spirit or essential features. you will be able to understand Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

10: Image Inspection System
100: user terminal
200: image inspection server
210: image collection unit
211: input unit 213: search unit
215: storage
220: image analysis unit
221: feature point extraction unit 223: image alignment unit
225: type classification unit 227: image preprocessing unit
229: text recognition unit
230: image editing unit
240: communication department
300: commerce server

Claims (18)

As a method for the image inspection server to inspect the food advertisement image,
collecting food advertisement images published on a commerce site;
comparing the collected food advertisement image with the original advertisement image, and searching for changes in the collected food advertisement image; and
determining whether to modify the collected food advertisement images according to the search results;
Food advertisement image inspection method comprising a. According to claim 1,
The step of collecting the food advertisement image is,
inputting at least the name of the food and the original advertisement image among the name of the food that needs to be inspected, the filtering phrase, and the original advertisement image;
searching a web page corresponding to the name and filtering phrase of the food; and
excluding unnecessary images from the searched web page and storing the corresponding food advertisement image;
Food advertisement image inspection method further comprising. According to claim 1,
The step of retrieving the change is:
extracting a reference point of the collected food advertisement image based on the original advertisement image, and designating an inspection area within the collected food advertisement image by using the extracted reference point;
Food advertisement image inspection method further comprising. 4. The method of claim 3,
After the step of designating the inspection area,
confirming a comparison type of the inspection area; and
pre-processing the inspection area according to the identified comparison type;
Food advertisement image inspection method further comprising. 4. The method of claim 3,
The step of designating the inspection area is
dividing the original advertisement image and the inspection area into a plurality of identical sections;
Food advertisement image inspection method further comprising. 5. The method of claim 4,
The step of confirming the comparison type is,
recognizing at least one image classification item among a background color, an image outline, and a character within the original advertisement image or the inspection area, and grouping one object in the recognized image classification item; and
determining whether one grouped object is a character, and classifying the comparison type for the inspection area into a character type or a non-character type;
Food advertisement image inspection method further comprising. 5. The method of claim 4,
The step of confirming the comparison type is,
extracting feature points from each of the original advertisement image and the inspection area when the comparison type is not a text type; and
searching for changes in the inspection area based on each feature point;
Food advertisement image inspection method further comprising. 5. The method of claim 4,
The step of confirming the comparison type is,
when the comparison type is a text type, recognizing a word in each of the original advertisement image and the inspection area, and searching for a change;
Food advertisement image inspection method further comprising. According to claim 1,
The step of determining whether to modify the
generating a modified food advertisement image by displaying the change in the collected food advertisement image when the change is found, and transmitting the modified food advertisement image to a commerce server managing the commerce site;
Food advertisement image inspection method further comprising. an image collection unit for collecting food advertisement images published on a commerce site;
an image analysis unit that compares the collected food advertisement image with the original advertisement image and searches for changes in the collected food advertisement image; and
an image editing unit that determines whether to correct the collected food advertisement images according to a search result;
Image inspection server that includes. 11. The method of claim 10,
The image collection unit,
an input unit for receiving at least the name of the food and the original advertisement image among the name of the food that needs to be inspected, the filtering phrase and the original advertisement image;
a search unit for searching a web page corresponding to the name and filtering phrase of the food in the commerce site; and
a storage unit for storing food advertisement images of the corresponding food by excluding unnecessary images from the searched web page;
Image inspection server further comprising a. 11. The method of claim 10,
The image analysis unit,
a feature point extraction unit for extracting a reference point of the collected food advertisement image based on the original advertisement image; and
an image alignment unit for designating an inspection area within the collected food advertisement image by using the extracted reference point;
Image inspection server further comprising a. 13. The method of claim 12,
a type classification unit for confirming a comparison type of the inspection area; and
an image pre-processing unit for pre-processing the inspection area according to the confirmed comparison type;
Image inspection server further comprising a. 13. The method of claim 12,
The image alignment unit,
dividing the original advertisement image and the inspection area into the same plurality of sections,
Image review server. 14. The method of claim 13,
The type classification unit,
Recognizes at least one image classification item among the background color, image outline, and text within the original advertisement image or the inspection area, and groups one object in the recognized image image classification item to determine whether a grouped object is a character and classifying the comparison type for the inspection area into a character type or a non-character type,
Image review server. 14. The method of claim 13,
The feature point extraction unit,
When the type classified by the type classification unit is not a text type, a feature point is extracted from each of the original advertisement image and the inspection area,
The image analysis unit,
Searching for changes in the inspection area based on each feature point,
Image review server. 14. The method of claim 13,
The image analysis unit,
a text recognition unit for recognizing words in each of the original advertisement image and the inspection area when the type classified by the type classification unit is a text type; further comprising,
An image inspection server that searches for changes in the words recognized by the text recognition unit. 14. The method of claim 13,
The image editing unit,
When the change is found, a modified food advertisement image is generated by displaying the change in the collected food advertisement image,
a communication unit transmitting the modified food advertisement image to a commerce server managing the commerce site;
Image inspection server further comprising a.

Images