JP2016162103A - Image processing program, image processing apparatus, and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a person accurately with a few collation patterns, regardless of face direction or hairstyle.SOLUTION: In an image processing apparatus 14, a skin color pixel detection section 24 detects skin-color pixels from an image corresponding to a video acquired from an imaging apparatus 12. A skin area generation section 26 generates a skin area, on the basis of the detected skin-color pixels. The skin color pixel detection section 24 and the skin area generation section 26 specify a skin area of a person from an image including the person. A color conversion section 28 converts a color of the skin area to a color of a hair area (hair color) so that the specified skin area may be identified with the hair area of the person by collation processing. A collation section 30 executes collation processing of collating an image of an object area with converted color with a detection mask, to detect a face of a person (face image), on the basis of a luminance value.SELECTED DRAWING: Figure 2

Description

  The disclosed technology relates to an image processing program, an image processing apparatus, and an image processing method.

  In recent years, it has been performed to detect a person appearing in an image captured by a monitoring camera or the like installed in various places such as an office, a shopping center, and a shopping street.

  As a technique for detecting a person, a technique for detecting a person by collating a collation pattern for detecting a person with an image is known.

  In addition, there is known a technique for detecting a person from an image obtained from an image captured by an imaging apparatus based on skin color or hair color.

JP-A-2005-157906 JP 2005-149143 A

  When a person is detected by collating a collation pattern with an image, a plurality of types of collation patterns may be required depending on the face orientation, hairstyle, and the like of the person shown in the video. When collating a plurality of types of collation patterns with images, for example, problems such as enormous time required for collation occur.

  One aspect of the disclosed technology is to accurately detect a person with a small number of collation patterns regardless of the face orientation and hairstyle.

  In one aspect, the skin area of the person is specified from an image in which the person is captured. In addition, at least one pixel of the skin region and the hair portion region so that the identified skin region and a hair portion region that can be regarded as the head portion of the person from the skin region are regarded as the same region in the matching process Convert information values. Furthermore, a process including performing the collation process for collating an image including the region whose value has been converted with a collation pattern for determining whether or not it is the head is executed.

  As one aspect, there is an effect that a person can be accurately detected with a small number of collation patterns regardless of the face orientation and hairstyle.

It is a schematic structure figure of an example of a person detection system concerning this embodiment. It is a functional block diagram showing the functional schematic structure of an example of the image processing apparatus which concerns on this embodiment. It is a block diagram which shows schematic structure of an example of the computer which functions as an image processing apparatus which concerns on this embodiment. It is a flowchart which shows an example of the image process performed with the image processing apparatus which concerns on this embodiment. It is a schematic diagram of an example of the image which the skin color pixel detection part of this embodiment converted. It is a schematic diagram of an example of the skin area | region produced | generated in the image shown in FIG. It is explanatory drawing for demonstrating the color conversion which converts the color of the pixel of a skin area | region into the color of a hair part area | region. It is explanatory drawing for demonstrating the color conversion which converts the color of a skin area | region into an achromatic color when there is no hair part area | region. It is a figure of an example of the mask for a detection of this embodiment. FIG. 8 is an explanatory diagram for explaining color conversion for converting the color of the hair region into the color of the skin region for the person shown in FIG. 7.

  Hereinafter, an example of the disclosed technique will be described in detail with reference to the drawings.

  FIG. 1 is a schematic configuration diagram of an example of an image processing system 10 according to the present embodiment. As shown in FIG. 1, the image processing system 10 includes an imaging device 12 and an image processing device 14. The imaging device 12 and the image processing device 14 are connected via a network 16. Note that the number of imaging devices 12 included in the image processing system 10 is not limited to this embodiment, and may be two or more.

  The image processing system 10 illustrated in FIG. 1 has a function of detecting a person in an image captured by the imaging device 12.

  The imaging device 12 has a function of capturing an image of a predetermined area, and is a so-called monitoring camera. The imaging device 12 of the present embodiment always captures a moving image in the monitoring area, and the captured video (moving image) is output to the image processing device 14 via the network 16.

  The image processing device 14 has a function of detecting a person (specifically, an image representing a person) in a video imaged by the imaging device 12.

  FIG. 2 is a functional block diagram showing a functional schematic configuration of an example of the image processing apparatus 14 of the present embodiment.

  As shown in FIG. 2, the image processing apparatus 14 includes a video acquisition unit 20, a video DB (Data Base) 22, a skin color pixel detection unit 24, a skin region generation unit 26, a color conversion unit 28, a collation unit 30, and a detection mask. DB32, detection result DB34, and the display part 36 are provided.

  The video acquisition unit 20 has a function of acquiring video captured by the imaging device 12 from the imaging device 12. The video acquired by the video acquisition unit 20 is stored in the video DB 22.

  The skin color pixel detection unit 24 is an example of a specifying unit of the disclosed technology, converts a video that is a moving image captured by the imaging device 12 into an image (still image) for each frame, and then converts skin color pixels from the converted image. It has a function to detect.

  The skin region generation unit 26 is an example of a specifying unit of the disclosed technology, and has a function of forming a skin region based on the skin color pixels detected by the skin color pixel detection unit 24. In the present embodiment, the skin area refers to the skin area (that is, the face area) of the head of the person shown in the image.

  The color conversion unit 28 is an example of a conversion unit of the disclosed technology, and has a function of converting the color of the skin region generated by the skin region generation unit 26. The color conversion unit 28 of the present embodiment uses a skin region and a region that can be regarded as a hair portion of a person shown in an image (hereinafter referred to as a “hair portion region”) as a target region, and changes the color of the target region to the same color. Convert. In the present embodiment, as a specific example, the color conversion unit 28 has a function of converting the color of the target region into the color of the hair region (hair) or an achromatic color. In the present embodiment, the color of the pixel corresponds to an example of pixel information of the disclosed technology.

  The collation unit 30 collates an image including the target region whose color has been converted by the color conversion unit 28 with a detection mask stored in the detection mask DB, thereby to detect a person (person's head) included in the image. ). The detection result of the collation unit 30 is stored in the detection result DB 34.

  The display unit 36 has a function of displaying information about the detected person.

  The image processing apparatus 14 can be realized by a computer 50 shown in FIG. 3, for example. The computer 50 includes a central processing unit (CPU) 52, an input / output interface (I / F) 54, a network I / F 56, a memory 58, and a nonvolatile storage unit 60. The CPU 52, input / output I / F 54, network I / F 56, memory 58, and storage unit 60 are connected to each other via a bus 80. The network I / F 56 is connected to the imaging device 12 via the network 16.

  The storage unit 60 can be realized by an HDD (Hard Disk Drive), a flash memory, or the like. An image processing program 62 for causing the computer 50 to function as the image processing apparatus 14 is stored in the storage unit 60 as a storage medium. The CPU 52 reads the image processing program 62 from the storage unit 60 and develops it in the memory 58, and sequentially executes each process of the image processing program 62.

  The image processing program 62 includes a video acquisition process 64, a skin color pixel detection process 66, a skin area generation process 68, a color conversion process 70, and a matching process 72.

  The video DB storage area 74 of the storage unit 60 functions as a storage area in which the video DB 22 is stored. Further, the detection mask storage area 76 of the storage unit 60 functions as a storage area in which the detection mask DB 32 is stored. Further, the detection result DB storage area 78 of the storage unit 60 functions as a storage area in which the detection result DB 34 is stored.

  The CPU 52 operates as the video acquisition unit 20 by executing the video acquisition process 64. The CPU 52 operates as the flesh color pixel detection unit 24 by executing the flesh color pixel detection process 66. Further, the CPU 52 operates as the skin region generation unit 26 by executing the skin region generation process 68. Further, the CPU 52 operates as the color conversion unit 28 by executing the color conversion process 70. Further, the CPU 52 operates as the matching unit 30 by executing the matching process 72.

  Thereby, the computer 50 that has executed the image processing program 62 functions as the image processing apparatus 14.

  The computer 50 is not limited to a so-called desktop personal computer. The computer 50 may be a laptop personal computer, a PDA (Personal Digital Assistants: portable information terminal device) represented by a tablet terminal or a smartphone, or the like.

  Note that the image processing apparatus 14 can be realized by, for example, a semiconductor integrated circuit, more specifically, an ASIC (Application Specific Integrated Circuit) or the like.

  Next, the operation of the image processing apparatus 14 according to this embodiment will be described.

  The video acquisition unit 20 of the image processing device 14 acquires the video captured by the imaging device 12 via the network 16 and stores it in the video DB 22 in advance before performing image processing. The timing at which the video acquisition unit 20 acquires video and stores it in the video DB 22 is not particularly limited. Examples of such timing include timing at predetermined time intervals such as every 10 minutes and every several hours. Further, for example, the captured image may be acquired in real time according to the timing when the imaging device 12 captures an image.

  In the image processing apparatus 14 according to the present embodiment, image processing is executed in a state where video is stored in the video DB 22. Image processing performed by the image processing apparatus 14 of the present embodiment will be described. FIG. 4 shows a flowchart of an example of image processing performed by the image processing apparatus 14 of the present embodiment.

  In the image processing apparatus 14 of the present embodiment, the image processing shown in FIG. 4 is executed at a predetermined timing or a timing instructed by a user who desires to detect a person from an image. The predetermined timing is not particularly limited, and examples thereof include a timing at every predetermined time such as every 10 minutes or every several hours. Further, for example, instead of every predetermined time, the video acquisition unit 20 may acquire the video from the imaging device 12 and store the video in the video DB 22.

  In step S100, the flesh color pixel detection unit 24 detects flesh color pixels from the video (image). Specifically, the flesh color pixel detection unit 24 reads a video from the video DB 22 and converts it into an image (still image) for each frame. In FIG. 5, the schematic diagram of an example of the image which the skin color pixel detection part 24 of this embodiment converted is shown. FIG. 5 shows an image 100 in which five persons (persons 102, 104, 106, 108, 110) are shown. The persons 102, 104, and 106 have faces and hairs in the image 100. The person 108 has a face in the image 100 but no hair. Further, the person 110 has a back head in the image 100 and no face. Hereinafter, as a specific example, a case where a person shown in the schematic diagram shown in FIG. 5 is detected will be described.

  The skin color pixel detection unit 24 detects skin color pixels included in the image. By detecting skin-colored pixels, pixel candidates that constitute a human face are detected.

  The method by which the flesh color pixel detection unit 24 detects flesh color pixels is not particularly limited. Is the color (R (Red), G (Green), B (Blue)) of each pixel in the image a predefined flesh color? What is necessary is just to determine. Moreover, you may use the method described in the following literature.

  Satoshi Matsuhashi, Kenji Fujimoto, Nobu Nakamura, Satoshi Minami, “Proposal of a modified HSV color system effective for face region extraction,” TV magazine, vol.49, no.6, pp.787-797, June 1995.

  Satoshi Matsuhashi, Osamu Nakamura and Toshi Minami, “Human-face extraction using modified HSV color system and personal identification through facial image based on isodensity maps,”. Proc. Canadian Conference on Electrical and Computer Engineering, pp. 909-912, 1995.

  In the next step S102, the skin region generation unit 26 groups the detected skin color pixels to generate a skin region. Specifically, by grouping and labeling adjacent skin-colored pixels, a candidate for a region corresponding to a face portion of a person appearing in an image is generated.

The method by which the skin region generation unit 26 generates the skin region is not particularly limited, and for example, a method described in the following document may be used.
Makoto Nagao Digital image processing pp.360-361.
RM Haralick and L. Shapiro, Computer and robot vision, Addison-Wesley, pp. 28-48, 1992.

  FIG. 6 shows a schematic diagram of an example of a skin region generated in the image 100 shown in FIG. As indicated by dotted lines in FIG. 6, four skin areas of persons 102, 104, 106, and 108 are generated from the image 100. For the person 110, no face is shown in the image 100, and no skin color pixels are detected, so no skin region is generated.

  In the next step S104, the color conversion unit 28 sets n = 1 for the skin region (n), and proceeds to the next step S106.

  In step S <b> 106, the color conversion unit 28 determines whether or not the hair part region exists at a position having a predetermined positional relationship with the generated skin region.

  The method by which the color conversion unit 28 determines whether or not there is a hair part region is not particularly limited. For example, when the skin region generated by the skin region generation unit 26 is regarded as a person's face, and there is an achromatic pixel at a position corresponding to the top of the head adjacent to the skin region (upper side in FIG. 6), A region formed by achromatic pixels may be determined as a hair region. In general, since the hair is often achromatic, the region formed by the achromatic pixel can be regarded as the hair portion region by regarding the achromatic color as the color of the hair. In this case, the position corresponding to the top of the head may be obtained in advance from the position of the top of the head relative to the skin region in the image 100 depending on the orientation of the imaging device 12 or the like.

  Note that the number of “positions corresponding to the top of the head adjacent to the skin region” may be one, but is preferably a plurality. Note that the number may be determined according to the size of the skin region, the size of the entire image, the resolution of the image, and the like.

  In addition, although the hair is not reflected in the position corresponding to the top of the head, the hair may be reflected in the position corresponding to the side of the face. Therefore, when determining the hair region from the region adjacent to the skin region in this way, the presence or absence of achromatic pixels other than the position corresponding to the top of the head is determined. The region formed by the colored pixels may be determined as the hair region.

  As another determination method, a moving area (foreground) is extracted from the difference from the background, and when the extracted foreground (person image) is in a position corresponding to the top of the skin area, the foreground is The hair region may be determined. In addition, the method of extracting the area | region with a motion from a background difference is not specifically limited, You may use the method described in the following literature.

  Moving object detection by time correlation type background judgment method, IEICE Transactions, D-II, vol.79, no.4, pp.568-576, 1995.

  If it is determined that there is hair, an affirmative determination is made and the process proceeds to step S108. From the image 100 shown in FIG. 5, the color conversion unit 28 determines that there are hair region for three persons 102, 104, and 106.

  In step S108, the color conversion unit 28 converts the color of the skin region and the color of the hair region into the same color. The color conversion unit 28 of this embodiment converts the color of the skin region and the head region (target region) into the same color by converting the color of the pixel of the hair region into the color of the pixel of the skin region.

  FIG. 7 is an explanatory diagram for explaining color conversion for converting the color of the pixel in the skin region into the color of the hair region. FIG. 7 shows a state in which the color of the skin region is converted into the color of the hair region for three persons 102, 104, and 106 that the color conversion unit 28 has determined to have hair.

  The method by which the color conversion unit 28 converts the color of the pixel in the skin area is not particularly limited. For example, as described above, when the achromatic region at the position corresponding to the top of the head is determined as the hair region, the color of the pixel in the achromatic region is changed to the color of the pixel in the skin region for each skin region. Convert. In this case, when a pixel of a plurality of colors is included in the achromatic color region, for example, it may be converted to the color of the largest number of pixels (achromatic color), or a predetermined one of the plurality of achromatic colors may be determined in advance. You may convert into the achromatic color (or the achromatic color close | similar to a predetermined achromatic color) determined.

  For example, as described above, when the hair region is determined from the background difference, the color of the pixel in the skin region is converted into the color of the hair region for each skin region. In this case, for example, as described above, when a plurality of color pixels are included in the hair region, it may be converted to a color corresponding to the most pixels, or a color corresponding to the boundary with the skin region May be converted to

  In the next step S110, the collation unit 30 determines whether or not the converted color is an achromatic color. If the color is achromatic, the determination is affirmative, and the process proceeds to step S116. On the other hand, if it is not an achromatic color, a negative determination is made and the process proceeds to step S112.

  In step S112, the collation unit 30 converts the converted color into an achromatic color. Here, the achromatic color to be converted by the collation unit 30 is not particularly limited, and examples thereof include a predetermined achromatic color. The predetermined achromatic color is preferably a color separated from the background color (for example, a distance on the color coordinate is a predetermined value or more).

  When the color conversion unit 28 determines that the achromatic region at the position corresponding to the top of the head is the hair region, the color to be converted in step S108 is an achromatic color, so the processing in steps S110 and S112 is omitted. May be. On the other hand, when the color conversion unit 28 determines the hair region from the background difference, the color to be converted in step S108 is not necessarily an achromatic color, and thus the processing in steps S108 and S112 is performed as in the main image processing. .

  On the other hand, if the color conversion unit 28 determines in step S106 that there is no hair region, the determination is negative and the process proceeds to step S114. The color conversion unit 28 determines from the image 100 shown in FIG. 5 that there is no hair region for one of the persons 108.

  In step S114, the color conversion unit 28 converts the color of the skin region to an achromatic color, and then proceeds to step S116. The achromatic color converted by the color conversion unit 28 is not particularly limited. For example, it may be converted to a predetermined achromatic color. The predetermined achromatic color is preferably a color separated from the background color (for example, the distance on the color coordinates is a predetermined value or more), as in step S112. The predetermined achromatic color in this step may be the same as or different from the predetermined achromatic color converted in step S112.

  Further, for example, the color may be converted to an achromatic color farthest from the color of the pixel corresponding to the background portion. The conversion method in this case will be described in detail with reference to FIG. FIG. 8 is an explanatory diagram for explaining color conversion for converting the color of the skin region into an achromatic color when there is no hair region.

  From the pixel (gx, gy) at the position of the center of gravity of the skin region, it proceeds in the vertical direction and the left-right direction (the arrow direction in FIG. 8) on the image, and is the most from the color (ri, gi, bi) of the pixel in contact with the skin region Convert to separated achromatic colors (c, c, c). “C” that defines the achromatic color may be derived when “J” in the following equation (1) is the maximum value. As shown in the following formula (1), “i” is the number of pixel positions used for deriving an achromatic color (“c”). For example, when using one pixel from each of the vertical direction and the horizontal direction to derive the most distant achromatic color with respect to the color of each pixel, i = 4.

... (1)
In the next step S116, the collation unit 30 increments n to n = n + 1, and proceeds to the next step S118.

  In step S118, the collation unit 30 determines whether n is larger than the number of detected skin regions. In the case of the image 100, since the number of detected skin regions is 4, it is determined whether or not n> 4. When it becomes negative determination, it returns to step S106 and repeats the process of said step S106-S116. On the other hand, when it becomes affirmation determination, it transfers to step S120.

  In step S120, the collation unit 30 collates the image including the target region with the detection mask stored in the detection DB 32, and then ends the image processing. The detection mask of the present embodiment is an example of a matching pattern according to the disclosed technique, and is a head shape model. Specifically, the detection mask of the present embodiment is a detection mask for detecting an image of a person's occipital region. As described above, in the present embodiment, since the color of the skin region that is the region corresponding to the face is converted into the color of the hair region, it is possible to perform collation using a detection mask for detecting the back of the head.

  FIG. 9 shows an example of the detection mask of this embodiment. The detection mask 120 shown in FIG. 9 is an example of a detection mask when the hair color is assumed to be black, the head portion color is the darkest, the upper body portion color is the second darkest, This is a detection mask assuming that the color of the part (background part) is the brightest. The head portion 122 of the detection mask 120 has a luminance value of −128, a luminance value of the upper body portion of 1, and a luminance value of the background portion 124 of 127.

  Although the collation method is not particularly limited, as a specific example, the collation unit 30 of the present embodiment moves the detection mask 120 over the entire image 100, and for each of the moved positions, the corresponding pixels in the region of the image 100. Check against the detection mask. First, the video acquisition unit 20 subtracts 128 from the luminance value represented by 0 to 255 for each pixel of the image 100, and converts the luminance value of each pixel to -128 to 127. Then, the luminance value of the detection mask 120 is multiplied by the luminance value of the pixel of the image 100 at the corresponding position, and the multiplied value is added over the entire detection mask 120 to derive an added value. When the image including the target region and the detection mask 120 are collated, the value of the added value is larger than when the image not including the target region and the detection mask 120 are collated. Therefore, in the image processing apparatus 14 according to the present embodiment, a threshold value for collation is obtained in advance through experiments or the like, and when the added value is equal to or greater than the threshold value, it is determined that the image is in the target region. Detect the person's image.

  Note that the collation unit 30 of the present embodiment differs in the size of the person's face shown in the image 100 depending on the position of the person, etc. In this embodiment, the size of the detection mask 120 is varied, The image 100 and the detection mask 120 are collated a plurality of times.

  Moreover, although the specific example of the detection mask when the hair is black is shown as the detection mask 120, it is possible to detect a person with white hair by separately using a detection mask when the hair is white. Can do. Thus, it is preferable that the detection mask is stored in advance in the detection mask DB 32 in accordance with the color of the hair.

  In this embodiment, since the detection mask for detecting the occipital image is used for collation, both the person whose face is reflected in the image and the person whose face is not reflected and whose occipital area is reflected are detected. can do. In the image 100 shown in FIG. 5, five persons 102, 104, 106, 108 with a face and a person 110 with a back head are detected.

  The collation unit 30 of the present embodiment stores the number of persons in the image in the detection result DB 34 as the detection result. Note that how the detection result is stored is not particularly limited, and the number of persons in the image may be stored in association with the information specifying the detection source image or video. In addition to the number of persons, information indicating the position of the detected person image may be stored. Moreover, the collation part 30 may display a detection result on the display part 36 as desired by the user or the like.

  In this way, by collating the detection mask 120 with the image 100, the human face and occipital images are detected from the skin area generated by the skin area generation unit 26, and the image of the area other than the face is detected. Not.

  As described above, in the image processing device 14 of the present embodiment, the flesh color pixel detection unit 24 detects flesh color pixels from an image corresponding to the video acquired from the imaging device 12. The skin region generation unit 26 generates a skin region based on the detected skin color pixels. Thereby, the skin color pixel detection unit 24 and the skin region generation unit 26 specify the skin region of the person from the image in which the person is captured. Then, the color conversion unit 28 sets the color of the skin region to the color of the hair region (the hair region) so that the identified skin region and the human hair region from the skin region are regarded as the same region in the matching process. Color). Furthermore, the collation unit 30 performs collation processing for collating the image of the target area whose color has been converted with the detection mask, and detects a human face (face image) based on the luminance value.

  In general, when detecting a human image using a detection mask, there is no need to prepare a detection mask for a large number of patterns or learn a large number of patterns according to the face orientation, hairstyle, etc. Must not.

  In addition, when the image from which the person is detected has a low resolution (specifically, the head of the person is 10 pixels or less), most of the pixels constituting the boundary between the face and the hair region are small. May constitute the boundary. Therefore, when a detection mask is obtained by learning the direction of the person's face, hairstyle, etc., the learning effect is low, and it becomes necessary to prepare detection masks for a large number of patterns.

  On the other hand, according to the image processing apparatus 14 of the present embodiment, since the color of the target region is the same color, the pattern for the occipital region having the least feature is used regardless of the direction of the person's face or the hairstyle. A person can be detected from the image using the detection mask. In addition, a person (person image) can be detected with high accuracy from a low-resolution image.

  Therefore, according to the image processing apparatus 14 of the present embodiment, it is possible to accurately detect a person with a small number of collation patterns regardless of the face orientation and hairstyle. In addition, in order to obtain a detection mask for detecting a person's face, it is not necessary to learn the direction of the person's face, the hairstyle, and the like.

  Moreover, according to the image processing apparatus 14 of this embodiment, the time required for collation can be shortened by reducing the number of collation patterns collated with the image.

  Further, according to the image processing apparatus 14 of the present embodiment, since the person is detected using the skin region and the hair region as the target region, the size of the region is larger than when only the skin region is used. Since (area) is increased, detection accuracy is improved.

  In the present embodiment, in step S108 of the image processing, the color of the skin region is converted to the color of the hair region, so that the color of the target region is converted to the same color. The method is not limited to this. For example, the color of the hair region may be converted to the color of the skin region. FIG. 10 is an explanatory diagram for explaining color conversion for converting the color of the hair region into the color of the skin region for the persons 102, 104, and 106 shown in FIG. As shown in FIG. 10, the colors of the target areas 103, 105, and 107 of the persons 102, 104, and 106 are converted into the colors of the skin areas, respectively. In general, since the color of a person's skin has less individual differences than the color of the hair, the number of detection masks can be reduced, so that the processing for matching can be performed at a higher speed. .

  When the color of the hair region is converted to the color of the skin region in this way, if the hair region is not detected as in the person 108, the detection mask is used without converting the color as it is. Can be detected. On the other hand, when the back of the head is shown like the person 110 and only the hair region is present, and the skin region is not detected, the detection may be performed separately using a detection mask for the back of the head. In addition, when converting the color of the hair region into the color of the skin region, the pixel of the color that can be regarded as the hair instead of detecting the skin color pixel in step S100 of the image processing and generating the skin region in step S102. May be detected to generate a hair region. In this case, the generated hair region and the skin region adjacent to the hair region may be converted into the same color as the target region.

  Note that the color of the skin region and the color of the hair region may be converted, for example, may be converted to the same achromatic color.

  In the present embodiment, the image including the target image and the detection mask are collated using the luminance of the pixel. However, saturation may be used instead of the luminance, or luminance and saturation may be used. Good.

  In the present embodiment, the mode in which the image processing program 62 is stored (installed) in advance in the storage unit 60 has been described. However, the image processing program 62 may be provided in a form recorded on a storage medium. . Examples of the storage medium include a CD-ROM (Compact Disc Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), and a USB (Universal Serial Bus) memory.

  Regarding the above embodiment, the following additional notes are disclosed.

(Appendix 1)
On the computer,
Identify the skin area of the person from the image of the person,
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. Convert the value,
Performing the collation process for collating an image including the region into which the value has been converted with a collation pattern for determining whether or not the head.
An image processing program for executing processing including the above.

(Appendix 2)
Converting the color information of the pixel information of the hair region into a value representing the color information of the skin region;
The image processing program according to attachment 1.

(Appendix 3)
Converting the color information of the pixel information of the skin region into a value representing the color information of the hair region,
The image processing program according to attachment 1.

(Appendix 4)
If the hair region does not exist,
The pixel information of the skin area is used as it is without being converted, or the pixel information of the skin area is converted into a predetermined value.
The image processing program according to any one of appendix 1 to appendix 3.

(Appendix 5)
The matching process performs matching using at least one of luminance and saturation of a pixel.
The image processing program according to any one of appendix 1 to appendix 4.

(Appendix 6)
A specific unit that identifies the skin area of the person from an image of the person;
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. A conversion unit for converting values;
A collation unit that performs the collation process for collating an image including the region in which the value is converted with a collation pattern for determining whether or not it is a head;
An image processing apparatus.

(Appendix 7)
The conversion unit converts color information of pixel information of the hair region into a value representing color information of the skin region.
The radiographic imaging apparatus according to appendix 6.

(Appendix 8)
The conversion unit converts color information of pixel information of the skin region into a value representing color information of the hair region;
The radiographic imaging apparatus according to appendix 6.

(Appendix 9)
If the hair region does not exist,
The conversion unit uses the pixel information of the skin region without conversion or converts the pixel information of the skin region into a predetermined value.
The radiographic imaging device according to any one of appendix 6 to appendix 8.

(Appendix 10)
The verification processing performed by the verification unit performs verification using at least one of luminance and saturation of a pixel.
The radiographic imaging device according to any one of appendix 6 to appendix 9.

(Appendix 11)
By computer
Identify the skin area of the person from the image of the person,
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. Convert the value,
Performing the collation process for collating an image including the region into which the value has been converted with a collation pattern for determining whether or not the head.
An image processing method.

(Appendix 12)
Converting the color information of the pixel information of the hair region into a value representing the color information of the skin region;
The image processing method according to attachment 11.

(Appendix 13)
Converting the color information of the pixel information of the skin region into a value representing the color information of the hair region,
The image processing method according to attachment 11.

(Appendix 14)
If the hair region does not exist,
The pixel information of the skin area is used as it is without being converted, or the pixel information of the skin area is converted into a predetermined value.
The image processing method according to any one of appendix 11 to appendix 13.

(Appendix 15)
The matching process performs matching using at least one of luminance and saturation of a pixel.
The image processing method according to any one of appendix 11 to appendix 14.

DESCRIPTION OF SYMBOLS 10 Person detection system 14 Image processing apparatus 24 Skin color pixel detection part 26 Skin area production | generation part 28 Color conversion part 30 Collation part 50 Computer 52 CPU
60 Storage Unit 62 Image Processing Program

Claims (7)

On the computer,
Identify the skin area of the person from the image of the person,
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. Convert the value,
Performing the collation process for collating an image including the region into which the value has been converted with a collation pattern for determining whether or not the head.
An image processing program for executing processing including the above. Converting the color information of the pixel information of the hair region into a value representing the color information of the skin region;
The image processing program according to claim 1. Converting the color information of the pixel information of the skin region into a value representing the color information of the hair region,
The image processing program according to claim 1. If the hair region does not exist,
The pixel information of the skin area is used as it is without being converted, or the pixel information of the skin area is converted into a predetermined value.
The image processing program according to any one of claims 1 to 3. The matching process performs matching using at least one of luminance and saturation of a pixel.
The image processing program according to any one of claims 1 to 4. A specific unit that identifies the skin area of the person from an image of the person;
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. A conversion unit for converting values;
A collation unit that performs the collation process for collating an image including the region in which the value is converted with a collation pattern for determining whether or not it is a head;
An image processing apparatus. By computer
Identify the skin area of the person from the image of the person,
The pixel information of at least one of the skin region and the hair region is such that the identified skin region and the hair region that can be regarded as the human hair from the skin region are regarded as the same region in the matching process. Convert the value,
Performing the collation process for collating an image including the region into which the value has been converted with a collation pattern for determining whether or not the head.
An image processing method.