JP2009252243A - Adult object determination method and system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an adult object determination method and system for determining a skin index from an image extracted from an object of adult object determination and determining whether the object is an adult object by using the skin index. <P>SOLUTION: The adult object determination method includes steps of: analyzing colors of an input image extracted from the object; acquiring an image of a color region of a flesh color system from the input image when the input image includes the color region of the flesh color system as a result of the analysis; determining a skin region by using the image of the color region; and determining whether the object is an adult object according to the skin index determined by using the area of the skin region. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to content determination, and more particularly, to a method and system for determining whether an object is an adult.

  With the development of popular media, adults are invading homes in earnest, and youths can easily access harmful adults through web pages or websites on the Internet, which are popular media. Is supposed to touch. In most homes, youths can easily connect to various websites using computers, and it is easy to access adults.

  Therefore, the Internet site provider performs an inspection process on various multimedia contents so that adults are not provided without a predetermined authentication process on a website operated by the Internet site provider.

  In particular, in the case of a search site provider or portal site provider, content such as URLs or images of adult sites, moving images, etc. so that they do not connect to a specified adult site or download adult content through a website operated by them. Warns that is associated with adult material, or blocks access to or access to hosts associated with adult sites.

  Then, the search site provider or portal site provider obtains images, moving images, etc. from the host, web page, website, etc. to be inspected, whether or not they are adult-oriented providers, and images, motion images obtained from them. Inspecting whether or not the image is an adult object, the image that is likely to be an adult object is first inspected, or by removing in advance based on the search results, the probability of exposing the adult object to the user is minimized There is a need to develop a method that can do this.

  However, the amount of inspection objects for judging adults is rapidly increasing, and it is difficult for the inspector to complete the inspection within the target time. Therefore, there is a demand for a method that enables the reexamination in addition to preventing the inspector's failure by conducting a quick and automatic inspection, but an effective method for judging adults for inspection is still presented. There is no current situation.

  In particular, in the case of content such as images and moving images, it is necessary to use image processing for adult object judgment, but there is a problem that the method and procedure for the image processing are unclear.

  Further, there is a need for a method that can prioritize the test object and dramatically increase the adult detection ratio with respect to time. An index (index) indicating the degree of adult object relevance is assigned to this priority. ) Is unclear how to determine.

  An object of the present invention is to solve the above-described problems, and an object of the present invention is to provide an adult product determination method and system capable of quickly testing an adult product by an automated method.

  Another object of the present invention is to provide a clear image processing method and procedure for inspection of an inspection object including contents such as images and moving images.

  Another object of the present invention is to provide a method and a procedure for setting an order of priority for determining an adult object and determining an index capable of expressing the degree of adult object relevance.

Means for solving the solution

  According to an aspect of the present invention for achieving the above object, there is provided a method for judging an adult object, the step of analyzing a color of an input image extracted from an object, and a color region of a skin color series as a result of the analysis. A skin color sequence image is acquired from the input image, a skin region is determined using the color region image, and a skin is determined using the area of the skin region. Determining whether or not the object is an adult according to an index.

  In order to achieve the above object, an adult object judgment system according to one aspect of the present invention includes a color analysis unit that analyzes the color of an input image extracted from an object, an analysis result of the color analysis unit, and the input image is a skin color series A skin color area determining unit that acquires an image of a skin color series color area from the input image, a skin area determining unit that determines a skin area using the image of the color area, and an area of the skin area And an adult object determining unit that determines whether or not the object is an adult object based on the skin index.

  According to the present invention described above, it is possible to provide an adult object determination method and system that can perform an inspection quickly and automatically even when the amount of an inspection object for determining an adult object is large. Become.

  Further, according to the present invention, an image for inspection is extracted from content such as an image acquired from a host, a website, a web page, or a moving image, and the image is analyzed to determine whether it is an adult product or not. It is possible to provide a clear image processing method and procedure.

  Also, according to the present invention, the priority of the test object can be determined by the skin index and the degree of the adult object relevance of the test object can be known, so whether or not the object is an adult object based on the skin index. It becomes possible to judge.

1 is a schematic block diagram of an adult product determination system according to an embodiment of the present invention. FIG. 5 is a block diagram schematically illustrating a skin region determination unit according to an embodiment of the present invention. It is a block diagram which shows roughly the adult thing judgment part by one Example of this invention. It is a figure which shows the image determined by the adult thing judgment system by one Example of this invention. 3 is a flowchart illustrating a method for determining an adult object according to an embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1 is a schematic block diagram of an adult product judgment system according to an embodiment of the present invention.

  An adult object determination system 102 according to an embodiment of the present invention includes a training unit 104, a color analysis unit 106, a skin color region determination unit 108, a skin region determination unit 110, and an adult object determination unit 112.

  The adult determination system 102 receives content such as images and moving images acquired from a website, a web page, a predetermined host, or the like as an inspection object for adult object determination. Further, the adult object determination system 102 may receive contents such as images and moving images stored in a predetermined database as inspection objects.

  The adult object determination system 102 determines a skin index for an image extracted from the inspection object by image processing, and determines whether the image is an adult object based on the skin index. Transmit to site operators. Thereafter, an operator such as a search site or a portal site determines whether or not to provide the user with the corresponding inspection object based on the received adult object determination result.

  The skin index represents the degree of adult object relevance of the inspection object, and the adult object determination system 102 can directly determine whether or not the inspection object is an adult object based on the skin index.

  In a modified embodiment, the adult product determination system 102 can use the skin index as a measure for determining the priority of adult product testing. For example, for an inspection object with a low skin index, it is determined that the object is not an adult object with a high priority of inspection, and an adult object determination result can be output so that this object is provided to the user. Therefore, operators such as search sites and portal sites will not re-inspect inspection objects having a low skin index thereafter.

  The training unit 104 performs training for determining a skin color range using a training set, and defines a skin color series according to the skin color range. Based on the skin color series defined by the training unit 104, the skin color area determination unit 108, which will be described later, determines whether or not an image extracted from the inspection object includes a color area of the skin color series.

  The training unit 104 determines a skin color range that defines a color of the skin color series by training the knowledge database using images included in the training set. Here, the training unit 104 determines the skin color range using the relative RGB of the trained image.

  Here, the relative RGB values are used to largely ignore the influence of luminance on the image by its definition. Therefore, if images having bright and dark images all have similar relative RGB values, it is determined that they belong to the same color region. That is, in the case of a color having a flesh color, it is determined that both a dark flesh color and a light flesh color belong to the same color region.

  This is for minimizing the influence of light and the influence of race upon determining the color area of the skin color series of the image. The relative RGB value of the image can be expressed by the following mathematical formula 1 using the RGB of the image.

Relative R = R / R + G + B;
Relative G = G / R + G + B;
Relative B = B / R + G + B;
(Mathematical formula 1)
The above mathematical formula 1 is a method of normalizing the difference between light and dark by dividing the R, G or B value by the R + G + B value. Since the R + G + B value is a characteristic that generally represents the brightness of a pixel, the influence on this can be minimized with relative RGB.

  Relative RGB values can be expressed by only two values depending on the definition. That is, if only the relative R and G values are known, the B value can be obtained from a predetermined calculation.

  In one embodiment of the present invention, an average and variance are obtained as a skin color range from the training set, and pixel values of the input image included in the average and variance are determined as skin pixels.

  The training unit 104 defines the relative RGB range of the skin color series from the relative RGB values of the images trained in the training set. In one embodiment, the relative RGB range of the flesh color series can be defined by the average and variance of the relative RGB values of the trained image.

  The color analysis unit 106 extracts an input image from the inspection object and analyzes the color. In the modified embodiment, the color analysis unit 106 can receive the input image extracted from the inspection object and analyze the color without directly extracting the input image.

  In one embodiment of the present invention, the inspection object is content such as an image or a moving image. Therefore, when the inspection object is an image, the input image can be the inspection object itself, and when the inspection object is a moving image, the input image is a still image extracted from the moving image by a method such as capture. Can be.

  The color analysis unit 106 acquires the color of each area constituting the input image through the analysis, and generates color information for the color. Here, the color analysis unit 106 generates relative RGB (relative RGB) of the input image pixels as color information.

  Also, the color analysis unit 106 determines whether the input image is a gray-series color region based on the color information that is the analysis result, and notifies the adult product determination unit 112 of the result. Thereafter, when receiving the notification from the color analysis unit 106 that the input image is composed of a gray color area, the adult thing determination unit 112 determines the skin index as a reference value. Details thereof will be described later.

  The skin color area determination unit 108 determines whether the input image includes a color area of the skin color series based on the color information that is the analysis result of the color analysis unit 106. When the input image includes a skin color series color area, a skin color series color area image is acquired from the input image. Here, the skin color area determination unit 108 determines whether the input image includes a skin color series color area using the skin color series defined by the training unit 104.

  In one embodiment, the skin color region determination unit 108 determines whether the relative RGB of the input image based on the analysis result of the color analysis unit 106 is included in the range of relative RGB of the skin color series defined by the training unit 104. By doing so, it can be determined whether or not a skin color series color region is included.

  In one embodiment of the present invention, the amount of data required to define the color range of the skin color series determined by the training unit 104 is obtained by using a relative RGB hue space that is relatively less affected by light. As a result, the color analysis unit 106 and the skin color region determination unit 108 reduce the amount of data and the calculation time required to determine whether or not the input image includes a skin color series color region.

  Specifically, the skin color region determination unit 108 receives the relative RGB values of the pixels of the input image from the color analysis unit 106, and receives the average and variance of the relative RGB values of the skin color series from the training unit 104. When the relative RGB values of the pixels of the input image are included in the color range based on the average and the variance, the pixel is determined as a skin color series color region.

  On the other hand, the skin color series color area image determined by the skin color area determination unit 108 may be an image obtained by removing a portion that is not a skin color series color area from the input image.

  In one embodiment, when the input image does not include a skin color series or gray series color area, the skin color area determination unit 108 notifies the adult matter determination unit 112 of the result, and based on this, the adult matter determination unit 112 It can be determined that the test object is not an adult.

  The skin area determination unit 110 determines a skin area using a color area image of the skin color series. Hereinafter, a specific configuration and operation of the skin region determination unit 110 will be described with reference to FIG.

  FIG. 2 is a block diagram schematically showing a skin region determination unit 110 according to an embodiment of the present invention. The skin region determination unit 110 includes a filter unit 202 and a connection region removal unit 204.

  The filter unit 202 filters the skin color series color region image determined by the skin color region determination unit 108 to remove noise. In one example, the filtering may be median filtering or average filtering.

  The connected region removing unit 204 removes a connected region having an area equal to or smaller than a set value from the noise-removed image output from the filter unit 202 to determine a skin region.

  Accordingly, in the skin area according to the embodiment of the present invention, noise is removed from the color area image of the skin color series, and the connected component system (connected component system is a general technique), and thus detailed description is omitted. This corresponds to an image in which a small connected area is removed by.

  The connected region removing unit 204 removes a small image region that does not greatly affect the measurement of the degree of adult object relevance by using a connected component. When calculating the index, the calculation time and complexity are reduced.

  In the embodiment described above, the connected area is removed after removing the noise from the color area image of the flesh color series. However, in the modified embodiment, after the connected area is removed by the connected area removing unit 204, the filter unit 202 is removed. The noise may be removed. Alternatively, the filter unit 202 may be omitted, and the connected region removal unit 204 may remove only the connected region from the skin color series color region image without performing the noise removal process. Alternatively, the connected region removing unit 204 may be omitted, and the skin region may be determined by removing only noise from the corresponding image without performing the connected region removing process.

  Referring to FIG. 1 again, the adult object determination unit 112 determines a skin index using the area of the skin region, and determines whether the object is an adult object based on the skin index. Hereinafter, the specific configuration and operation of the adult object determination unit 112 will be described with reference to FIG.

  FIG. 3 is a block diagram schematically showing the adult product determination unit 112 according to an embodiment of the present invention. The adult product determination unit 112 includes a skin index determination unit 302 and a skin index comparison unit 304.

  The skin index determination unit 302 determines the skin index using the area of the skin region. In one embodiment, the skin index determination unit 302 determines the skin index according to the ratio of the area of the skin region to the area of the input image extracted from the inspection object. In one embodiment, the skin index is represented by Equation 2 below.

  When the skin index determination unit 302 receives a notification from the color analysis unit 106 that the input image is made of a gray color area, the skin index determination unit 302 determines the skin index as a reference value. This is because if the input image does not include a color region of the skin color series, it is not possible to determine the skin index associated with the degree of human skin exposure, and thus determine the skin index as a predefined reference value. That is why. Details thereof will be described later.

  In addition, when the skin index is equal to or lower than the first critical value or equal to or higher than the second critical value, the skin index determination unit 302 determines the skin index as a basic value again. This is because if there are too few or too many skin regions, the input image is likely to be an image that does not have a human shape or a simple box image filled with skin-colored colors, so This is because it is determined that the degree of association is low.

  For example, in an adult object judgment criterion for judging an image having a skin index exceeding 0.5 as an adult object, 0.005 is extremely small for an image having a skin index of 0.01 or less or 0.99 or more. By assigning the skin index again, the image is not judged as an adult.

  The skin index comparison unit 304 determines whether or not the test object is an adult by comparing the skin index determined by the skin index determination unit 302 with a specific reference value. That is, the skin index comparison unit 304 determines whether or not the test object is an adult based on the size of the skin index.

  For example, when the skin index exceeds 0.5, the skin index comparison unit 304 determines that the inspection object is an adult object. Then, the result is provided to operators such as search sites and portal sites.

  On the other hand, when the skin image index of 0.5 is given because the input image is composed of a gray series color area, the skin index comparison unit 304 displays the text extracted from the inspection object or the web that is the source of the inspection object. It is determined whether the text extracted from the page or the like is an adult text after considering whether the text is already specified for adults.

  For example, when the input image is a gray series nude female image, the skin index determination unit 302 sets the skin index to 0.5. Since the input image is a nude female image, it must be determined as an adult. Therefore, in this case, the skin index comparison unit 304 determines that the text extracted from the inspection object or the text extracted from the inspection object source host, web page, website, bulletin board, BBS is “sex” “porn” ( When the text for adults such as “porno” is included, the object to be inspected is determined as an adult by weighting the skin index of 0.5 even though the corresponding skin index is 0.5.

  When the input image does not include a skin color series or gray series color region, the skin index comparison unit 304 is notified of the result from the skin color region determination unit 108. Then, the skin index comparison unit 304 determines that the test object is not an adult based on the notification.

  FIG. 4 is a diagram illustrating an image determined by the adult product determination system 102 according to an embodiment of the present invention.

  Referring to FIG. 4, (a) in FIG. 4 is an input image extracted from an inspection object, which is a female image in which a certain portion of skin is exposed.

  The image of FIG. 4A is analyzed by the color analysis unit 106, and the color analysis unit 106 outputs the color information of the corresponding image that is the analysis result.

  FIG. 4B shows a skin color sequence color region image generated by the skin color region determination unit 108 extracting the skin color sequence color region based on the color information.

  FIG. 4C shows an image after the noise 402 appearing in the color region image of the skin color series is removed by the filter unit 202 of the skin region determination unit 110.

  FIG. 4D is an image after the connected region 404 having an area equal to or smaller than the set value is removed by the connected region removing unit 202 of the skin region determining unit 110.

  FIG. 4E shows an image in which the skin region 406 is finally determined by the connected region removing unit 202 of the skin region determining unit 110. Based on the image shown in FIG. 4 (e), the adult object determination unit 112 determines the skin index and determines whether or not it is an adult object.

  FIG. 5 is a flowchart illustrating an adult product determination method according to an embodiment of the present invention.

  First, the color of the input image extracted from the inspection object is analyzed (S502). In one embodiment, the inspection object may include at least one of a web page, a website, a host, an image, a moving image, and a posting.

  Next, based on the analysis result, it is determined whether the input image is composed of a gray color area (S504). If the input image is composed of a gray color area, the reference value is determined as the skin index, and it is determined whether the text extracted from the source of the input image is already defined for adults (S524). If the text extracted from the source of the input image is text for adults, the test object is determined to be an adult by weighting the reference value (S520). If not, the test object is not an adult. Is determined (S522b).

  On the other hand, if the input image does not include a gray color area (S504) as a result of the analysis in S504, it is determined whether the input image includes a skin color color area (S506). If the input image does not include a skin color series color area, it is determined that the inspection target is not an adult (S522a). However, if the input image includes a skin color series color area, a skin color series color area image is acquired from the input image (S508).

  In one embodiment, the skin color series can be defined by a skin color range determined using a training set.

  In one embodiment, whether or not a skin color series color region is included is determined by determining whether or not the relative RGB of the input image based on the analysis result is included in a relative RGB range corresponding to the skin color series. It can be. Here, the relative RGB is as described in the mathematical formula 1, and a specific description thereof is omitted.

  Subsequently, the skin color series color region image is filtered to remove noise (S510). In one example, the filtering may be median filtering or average filtering.

  Next, an image from which the connected area having an area equal to or smaller than the set value is removed from the image from which noise has been removed is obtained (S512), and a skin color series color area of the image from which the connected area is removed is determined as a skin area. (S514).

  In the embodiment described above, the connected area is removed after removing the noise from the color area image of the flesh color series. However, in the modified embodiment, the noise may be removed after removing the connected area. Alternatively, only the connected region may be removed from the skin color series color region image without going through the noise removing step (S510), or only the noise may be removed from the corresponding image without going through the connected region removing step (S512). Also good.

  Next, the skin index is determined based on the ratio of the area of the skin area to the area of the input image (S516). In one embodiment, the skin index may be determined again as the base value if the skin index determined using the area of the skin area is less than or equal to the first critical value or greater than or equal to the second critical value.

  Next, it is determined whether the skin index is larger than a specific reference value (= 0.5) (S518). If the skin index is larger than the specific reference value (= 0.5), it is determined that the test object is an adult (S520).

  The adult determination method can be embodied in a program form performed using various computer means. In this case, the program for performing the adult object determination method is a hard disk, CD-ROM, DVD, ROM, RAM, or It is stored in a computer-readable recording medium such as a flash memory.

  Those skilled in the art to which the present invention pertains can understand that the present invention can be implemented in other specific forms without departing from the technical idea and essential features thereof.

  Accordingly, the embodiments described above are illustrative in any aspect and should not be understood as limiting. The scope of the present invention is defined by the terms of the claims, rather than the detailed description above, and any modifications or variations that are derived from the claims and their equivalents are intended to be within the scope of the present invention.

DESCRIPTION OF SYMBOLS 102 Adult judgment system 104 Training part 106 Color analysis part 108 Skin color area determination part 110 Skin area determination part 112 Adult determination part

Claims (22)

Analyzing the color of the input image extracted from the object;
As a result of the analysis, when the input image includes a skin color series color area, obtaining an image of the skin color series color area from the input image;
Determining a skin area using an image of the color area;
Determining whether the object is an adult by a skin index determined using the area of the skin region;
A method for judging an adult product, comprising: The skin index is
The method according to claim 1, wherein the method is determined by a ratio of an area of the skin region to an area of the input image. Determining whether the input image is a gray-series color region, as a result of the analysis;
When it is determined that the input image is composed of a gray color area, the skin index is determined as a reference value, and it is determined whether or not the text extracted from the object is already defined adult text. Thus, it is determined whether or not the object is an adult object. The adult object determination method according to claim 1, wherein: Determining the skin area comprises:
Filtering the image of the color region to remove noise;
Determining the skin region using the denoised image;
The method for judging an adult product according to claim 1, comprising: The filtering is
The method of determining an adult object according to claim 4, comprising median filtering or average filtering. Determining the skin area comprises:
Obtaining an image obtained by removing a connected component having an area equal to or smaller than a set value from the image of the color region;
Determining the skin area using the removed image of the connected area;
The method for judging an adult product according to claim 1, comprising: The skin color series is
The adult object judging method according to claim 1, wherein the adult object judging method is defined by a skin color range determined using a training set. The determination of whether to include the color area,
2. The method of determining an adult object according to claim 1, comprising determining whether relative RGB of the input image based on the analysis result is included in a range of relative RGB corresponding to the skin color series. . The relative RGB of the input image is
9. The adult object judging method according to claim 8, wherein each of the RGB values of the input image is a ratio of the RGB values of the input image.   The skin index is determined again as a basic value when the skin index determined using the area of the skin region is equal to or lower than a first critical value or equal to or higher than a second critical value. The adult product judgment method described in 1. The object is
The method for judging an adult object according to claim 1, comprising at least one of a web page, a website, a host, an image, a moving image, and a posting.   The computer-readable recording medium with which the program for performing the method of any one of Claims 1-11 was recorded. A color analyzer that analyzes the color of the input image extracted from the object;
As a result of analysis by the color analysis unit, when the input image includes a skin color series color area, a skin color area determination unit that acquires an image of the skin color series color area from the input image;
A skin region determination unit that determines a skin region using an image of the color region;
Determining an index of skin using the area of the skin region, and determining whether or not the object is an adult based on the skin index;
An adult product judgment system characterized by including: The adult matter determination unit
14. The adult product determination system according to claim 13, wherein the skin index is determined by a ratio of the area of the skin region to the area of the input image. The adult matter determination unit
As a result of the analysis by the color analysis unit, when the input image is composed of a gray color area, the skin index is determined as a reference value, and the text extracted from the object is an already defined adult text. 14. The adult product determination system according to claim 13, wherein it is determined whether or not the object is an adult by determining whether or not there is an object. The skin area determination unit
A filter unit for filtering the image of the color region to remove noise;
A skin region determination unit that determines the skin region using the image from which the noise has been removed;
The adult product judgment system according to claim 13, comprising: The filter unit is
The adult thing determination system according to claim 16, wherein filtering including intermediate value filtering or average value filtering is performed. The skin area determination unit
A connected region removing unit for removing a connected region having an area equal to or smaller than a set value from the color region image;
A skin area determination unit that determines the skin area using an image from which the connection area has been removed;
The adult product judgment system according to claim 13, comprising: A training unit that performs training for determining a skin color range using a training set, and further includes a training unit that defines the skin color series according to the skin color range determined by the training,
The adult matter determination system according to claim 13, wherein the skin color region determination unit determines whether or not the color region is included using a skin color series defined by the training unit. The skin color area determination unit
14. The method according to claim 13, wherein it is determined whether or not the color region is included according to whether or not the relative RGB of the input image based on the analysis result is included in a relative RGB range corresponding to the skin color series. Adult product judgment system as described in. The relative RGB of the input image is
21. The adult object determination system according to claim 20, wherein each of the RGB values of the input image is a ratio of the RGB values of the input image. The adult matter determination unit
14. The adult product determination system according to claim 13, wherein when the skin index is equal to or lower than the first critical value or equal to or higher than the second critical value, the skin index is determined again as a basic value.

Images