JP2016009203A - Wrinkle detection apparatus and wrinkle detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a wrinkle detection apparatus capable of precisely detecting wrinkles even when a skin is glossy.SOLUTION: A wrinkle detection apparatus 100 is an apparatus for detecting an area of wrinkles on a skin included in an image. The wrinkle detection apparatus 100 includes: an image acquisition section 120 that acquires an image; an edge emphasis processing section 220 that generates an unsharp image by smoothing an acquired image, and performs an edge emphasis processing based on the generated unsharp image to generate an edge emphasized image as an image in which edges of the image are emphasized; and a wrinkle detection processing section 230 that performs a wrinkle detection processing on the generated edge emphasized image. The edge emphasis processing is a processing to generate an edge emphasized image by reducing influences in an area where the brightness is higher in an image.

Description

  The present invention relates to a wrinkle detection device and a wrinkle detection method for detecting a wrinkle region of a skin included in an image.

  Conventionally, a wrinkle region of a skin is detected from an image obtained by photographing a face. Here, the wrinkle region is a linear image region of a wrinkle portion projected in an image.

  However, due to the influence of illumination or the like, the wrinkle region may not be detected, or a portion that is not a wrinkle may be detected as a wrinkle region. Therefore, for example, Patent Document 1 discloses a technique for improving the accuracy of detection of a wrinkle region from an image (hereinafter referred to as “wrinkle detection”).

  The technique described in Patent Document 1 (hereinafter referred to as “conventional technique”) applies a so-called unsharp masking technique. In the unsharp masking technique, a difference image indicating a difference between an image and an unsharp image that is an image obtained by smoothing the image is generated, and the generated difference image is combined with a base image. An image obtained by such processing is an edge-enhanced image in which the edge of the base image is enhanced. The prior art determines the flatness of the pixel values of each part of the image by comparing the average values of the pixel values for the two types of ranges, and the difference when performing the above synthesis according to the determined flatness Change the degree of image influence.

  By applying an edge detection filter to the edge-enhanced image and comparing the gradient value of each part of the image with a threshold value, a region with a strong edge can be detected. Here, the gradient value is a value indicating the degree of change of the pixel value in the image. Therefore, by applying the conventional technique to an image obtained by photographing the skin, it is possible to reduce the possibility that elements other than wrinkles such as spots are erroneously detected as wrinkles in a flat skin area without wrinkles. Detection can be performed.

JP-A-11-66300 Japanese Patent Application Laid-Open No. 2011-8643 JP 2011-115460 A JP-A-6-76062

  However, in the related art, when gloss is generated on the surface of the skin due to light hitting the skin, elements other than wrinkles may be detected as wrinkle regions (hereinafter referred to as “false detection”). This is because the contrast strength generated in the skin due to gloss may be at a level close to the contrast strength generated due to wrinkles.

  Therefore, it is conceivable to set the threshold value of the gradient value high (strict). However, the higher the threshold, the higher the possibility that an actually existing soot will not be detected (hereinafter referred to as “detection omission”).

  Therefore, even when the skin is glossy, it is desired that both false detection and detection omission can be reduced and high-precision wrinkle detection can be realized.

  An object of the present invention is to provide a wrinkle detection device and a wrinkle detection method that can perform wrinkle detection with high accuracy even when the skin is glossy.

  The wrinkle detection device of the present disclosure is a wrinkle detection device that detects a wrinkle region of skin included in an image, and generates an unsharp image by smoothing the acquired image and an image acquisition unit that acquires the image An edge enhancement processing unit that performs an edge enhancement process for generating an edge enhanced image that is an image in which an edge of the image is enhanced based on the generated unsharp image, and the generated edge enhanced image An edge detection processing unit that detects the eyelid region by performing edge detection processing, and the edge enhancement processing suppresses the influence of a portion with high luminance in the image, and It is a process to generate.

  The wrinkle detection method of the present disclosure is a wrinkle detection method for detecting a wrinkle region of skin included in an image, the step of acquiring the image, and smoothing the acquired image to generate an unsharp image, Based on the generated unsharp image, generating an edge enhanced image that is an image in which an edge of the image is enhanced, and performing edge enhancement processing on the generated edge enhanced image And performing the step of detecting the wrinkle region, and the edge enhancement process is a process of generating the edge enhanced image while suppressing an influence of a portion having a high luminance in the image.

  According to the present disclosure, high-precision wrinkle detection can be performed even when the skin is glossy.

The block diagram which shows an example of a structure of the wrinkle detection apparatus which concerns on Embodiment 1 of this invention. The block diagram which shows an example of a structure of the wrinkle detection apparatus which concerns on Embodiment 2 of this invention. The block diagram which shows an example of a structure of the skin state detection part in this Embodiment 2. FIG. The flowchart which shows an example of operation | movement of the wrinkle detection apparatus which concerns on this Embodiment 2. Figure showing the state of wrinkle detection by conventional unsharp masking The figure which shows the mode of the wrinkle detection by the wrinkle detection apparatus which concerns on this Embodiment 2.

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

(Embodiment 1)
Embodiment 1 of the present invention is an example of a basic aspect of the present invention.

<Configuration of soot detection device>
First, the configuration of the wrinkle detection device according to the present embodiment will be described.

  FIG. 1 is a block diagram illustrating an example of the configuration of the wrinkle detection device according to the present embodiment.

  In FIG. 1, wrinkle detection device 100 is a device that detects a wrinkle region of a skin included in an image, and includes image acquisition unit 120, edge enhancement processing unit 220, and wrinkle detection processing unit 230.

  The image acquisition unit 120 acquires an image.

  The edge enhancement processing unit 220 smoothes the acquired image to generate an unsharp image, and based on the generated unsharp image, generates an edge enhanced image that is an image in which the edge of the image is enhanced. Process. However, the edge enhancement process is a process for generating an edge enhanced image while suppressing the influence of a portion with high luminance in the image.

  The wrinkle detection processing unit 230 detects a wrinkle region by performing edge detection processing on the generated edge-enhanced image.

  The wrinkle detection device 100 includes, for example, a CPU (Central Processing Unit), a storage medium such as a ROM (Read Only Memory) storing a control program, and a working memory such as a RAM (Random Access Memory), although not shown. In this case, the function of each part of the eyelid detection device 100 is realized by the CPU executing the control program.

  Such wrinkle detection apparatus 100 generates an edge-enhanced image in which contrast caused by gloss generated on the surface of the skin (hereinafter simply referred to as “gloss”) is reduced by suppressing the influence of a portion with high luminance. can do. Thereby, the wrinkle detection apparatus 100 can strengthen the contrast generated due to the gloss in the edge-enhanced image relative to the contrast generated due to the wrinkle. Therefore, wrinkle detection apparatus 100 can perform wrinkle detection with high accuracy even when the skin is glossy.

(Embodiment 2)
The second embodiment of the present invention is an example of a specific mode when the present invention is applied to an apparatus that detects a wrinkle region of a facial skin and presents it to a user.

  FIG. 2 is a block diagram showing an example of the configuration of wrinkle detection apparatus 100 according to the present embodiment.

  2, the wrinkle detection device 100 includes a photographing unit 110, an image acquisition unit 120, a facial part acquisition unit 130, a skin condition detection unit 140, an image generation unit 150, and a display unit 160.

  The photographing unit 110 is, for example, a still camera including a lens and a color image sensor, and photographs an image of the user's face. Then, the imaging unit 110 outputs the captured image to the image acquisition unit 120. The operation of the imaging unit 110 is controlled by the image acquisition unit 120, for example.

  The image acquisition unit 120 performs necessary image quality processing such as brightness adjustment on the input image and outputs the input image to each of the facial part acquisition unit 130 and the image generation unit 150. As such image quality processing, processing suitable for the contents of edge enhancement processing and edge detection processing described later is employed. In the following description, an image output from the image acquisition unit 120 is referred to as a “captured image”.

  Note that the captured image may be reversed in the left-right direction in the image acquisition unit 120 or another device unit.

  The facial part acquisition unit 130 detects the position of the facial part in the captured image from the captured image. Here, the face parts are each part constituting the face such as eyes, nose and cheeks, and can be defined by the positions of the facial feature points such as the eyes. The facial part acquisition unit 130 detects the position of the facial part by extracting the facial feature points from the captured image using a known image feature point detection method such as pattern matching. Then, the facial part acquisition unit 130 outputs the captured image and facial part position information indicating the position of each detected facial part to the skin state detection unit 140.

  The skin condition detection unit 140 detects a wrinkle region in the captured image. Then, skin condition detection unit 140 outputs wrinkle region information indicating the detected wrinkle region to image generation unit 150.

  FIG. 3 is a block diagram illustrating an example of the configuration of the skin condition detection unit 140.

  In FIG. 3, the skin condition detection unit 140 includes a detection region determination unit 210, an edge enhancement processing unit 220, a wrinkle detection processing unit 230, and a spot detection unit 240.

  The detection area determination unit 210 determines a detection area that is a target for wrinkle detection based on the facial part position information input from the facial part acquisition unit 130. For example, the detection area determination unit 210 determines the area from the lower eyelid to the left cheek and the area from the lower eyelid to the right cheek as detection areas based on the positions of both eyes and the nose. Such a region is generally a region where wrinkles are likely to occur. Then, the detection region determination unit 210 outputs the captured image and detection region information indicating the determined detection region to the edge enhancement processing unit 220 and the spot detection unit 240.

  Note that the detection area determination unit 210 preferably excludes the area occupied by the eyelashes, the hair, the edge of the glasses, and the like from the image area. For example, the eyelash region can be removed by using a known image processing technique such as the technique described in Patent Document 2 or Patent Document 3.

  The edge enhancement processing unit 220 smoothes the captured image and generates an unsharp image. The edge enhancement processing unit 220 performs edge enhancement processing that generates an edge enhanced image that is an image in which the edge of the captured image is enhanced based on the generated unsharp image. However, the edge enhancement process is a process of generating an edge enhanced image while suppressing the influence of a portion with high luminance in the captured image. The edge enhancement processing unit 220 includes a gray scale calculation unit 221, a smoothing processing unit 222, a minimum value calculation unit 223, and a difference calculation unit 224.

  The gray scale calculation unit 221 performs a known gray scale conversion process on the input photographed image (for example, the photographed image 301 in FIG. 6), so that a gray scale image that does not include color information (for example, FIG. 6). Of the gray scale image 302). That is, the grayscale image is an image including only the luminance of each pixel of the captured image as a pixel value. Then, the gray scale calculation unit 221 outputs the captured image, the detection area information, and the generated gray scale image to the smoothing processing unit 222.

  The smoothing processing unit 222 performs a known smoothing process using a moving average filter, a Gaussian filter, or the like on the input grayscale image, so that an unsharp image (for example, FIG. An unsharp image 303) is generated. Then, the smoothing processing unit 222 outputs the grayscale image, the detection area information, and the generated unsharp image to the minimum value calculation unit 223.

  Note that, as disclosed in Patent Document 4, the smoothing processing unit 222 may generate two unsharp images by changing the smoothing strength (smoothing strength) of the Gaussian filter. Then, the smoothing processing unit 222 may output the two generated smoothed images to the minimum value calculating unit 223 instead of the grayscale image and the unsharp image. An unsharp image is obtained by smoothing fine noise components of a grayscale image. Therefore, in this case, the result of edge enhancement processing described later is improved.

  The minimum value calculation unit 223 synthesizes the input grayscale image and the input unsharp image in a state in which the influence of the high-luminance portion in these images is suppressed, thereby generating a high-intensity suppression image (for example, FIG. To generate a high luminance suppression image 310a).

  More specifically, the minimum value calculation unit 223 increases the pixel value of the pixel having the minimum luminance among the luminance of each pixel of the grayscale image and the luminance of the corresponding pixel in the unsharp image. The pixel value (luminance value) of the corresponding pixel in the luminance-suppressed image is used. Note that the minimum value calculation unit 223 determines a pixel value of the high-intensity suppression image corresponding to at least each pixel of the detection area indicated by the input detection area information. Then, the minimum value calculation unit 223 outputs the grayscale image, the unsharp image, the detection area information, and the generated high brightness suppression image to the difference calculation unit 224.

  When the above-described two unsharp images are input, the minimum value calculation unit 223 uses the pixel value of the pixel having the minimum luminance between the corresponding pixels in these two unsharp images as the high-luminance suppression image. Is the pixel value of the corresponding pixel.

  The difference calculation unit 224 generates a difference image (for example, the difference image 311a in FIG. 6) indicating the difference between the input high-intensity suppression image and the input unsharp image. That is, the difference calculation unit 224 sets a value obtained by subtracting the luminance value of the corresponding pixel in the unsharp image from the luminance value of each pixel of the high luminance suppression image as the luminance value of the corresponding pixel in the difference image. Further, the difference calculation unit 224 combines the input grayscale image and the generated difference image to generate an edge enhanced image (for example, the edge enhanced image 312a in FIG. 6). Then, the difference calculation unit 224 outputs the detection area information and the generated edge enhanced image to the wrinkle detection processing unit 230.

  In addition, when a high-intensity suppression image is produced | generated from the above-mentioned two unsharp image, the difference calculation part 224 shows the difference image which shows the difference between a high-intensity suppression image and an unsharp image with high smoothing intensity | strength. Generate. Then, the difference calculation unit 224 combines the input grayscale image and the difference image to generate an edge enhanced image.

  Further, the minimum value calculation unit 223 and the difference calculation unit 224 do not necessarily have to generate the high-intensity suppression image and the difference image. That is, the minimum value calculation unit 223 and the difference calculation unit 224 calculate the difference between the pixel value of the pixel having the minimum luminance described above and the pixel value of the pixel of the unsharp image, and the calculated difference And a value obtained by combining the pixel values of the pixels of the grayscale image. Then, the minimum value calculation unit 223 and the difference calculation unit 224 may output the calculated values to the wrinkle detection processing unit 230 in association with the pixel positions of the base captured image.

  The wrinkle detection processing unit 230 calculates the gradient value of each part of the edge-enhanced image by a known edge detection process using, for example, a Gabor filter, a Laplacian filter, a Prewitt filter, a Sobel filter, or the like. Then, the wrinkle detection processing unit 230 detects the wrinkle region from the grayscale image (that is, from the captured image) by comparing the calculated gradient value with a predetermined threshold value. Here, the gradient value is a value indicating the degree of change of the pixel value in the image, and is a value that is higher at a stronger edge portion.

  That is, the wrinkle detection processing unit 230 performs a known edge detection process. When the gradient value increases as the degree of change in the pixel value increases, a region where the gradient value is equal to or greater than the threshold value is detected as a heel region. Note that the wrinkle detection processing unit 230 detects a wrinkle region at least for the detection region indicated by the input detection region information.

  Note that the predetermined threshold value may be a fixed value or a value that changes according to the amount of wrinkle area detected or a value specified by the user. Then, wrinkle detection processing section 230 outputs wrinkle area information indicating the detected wrinkle area to image generation section 150 (see FIG. 2).

  The spot detection unit 240 detects a skin spot area included in the shot image from the input shot image. For example, the spot detection unit 240 extracts a pixel having a value equal to or less than a threshold value using at least RGB signals for each region indicated by the input region position information from the captured image. Thus, the spot area detection is performed. Then, the spot detection unit 240 outputs the spot area information indicating the detected spot area to the image generation unit 150.

  The image generation unit 150 in FIG. 2 generates a wrinkle image indicating a wrinkle region in the captured image based on the input wrinkle region information, and a wrinkle-enhanced image obtained by superimposing the generated wrinkle image on the captured image (for example, FIG. 6 皺 emphasized image 313a) is generated. The wrinkle image is, for example, an image showing the position of a wrinkle line in which the wrinkle region is filled with a predetermined color. Then, the image generation unit 150 outputs the generated wrinkle enhanced image to the display unit 160.

  Note that the image generation unit 150 may use only a part of the plurality of eyelid areas indicated by the eyelid area information as targets for generating the eyelid images. For example, the image generation unit 150 may generate only the eyelid regions that fall within the range of 10% of the area of all the eyelid regions indicated by the eyelid region information in descending order of the area of the continuous regions. Good. Further, the wrinkle-enhanced image may be reversed in the left-right direction in the image generation unit 150 or the display unit 160 at the subsequent stage. Further, the image generation unit 150 may generate a spot image indicating a spot area in the captured image based on the spot area information, and include the generated haze image in the haze-enhanced image.

  The display unit 160 includes, for example, a liquid crystal display with a touch panel, and displays the input wrinkle emphasized image.

  Although not shown, the wrinkle detection device 100 includes, for example, a CPU, a storage medium such as a ROM storing a control program, and a working memory such as a RAM. In this case, the function of each part of the eyelid detection device 100 is realized by the CPU executing the control program.

  The wrinkle detection apparatus 100 having such a configuration can generate an edge-enhanced image while suppressing the influence of a portion with high luminance, and can perform wrinkle detection using the edge-enhanced image.

<Operation of soot detection device>
Next, the operation of wrinkle detection device 100 will be described.

  FIG. 4 is a flowchart illustrating an example of the operation of the eyelid detection apparatus 100.

  In step S <b> 1100, the image acquisition unit 120 uses the imaging unit 110 to acquire a captured image of the user's face.

  In step S1200, the facial part acquisition unit 130 detects the position of each facial part from the captured image.

  In step S1300, the detection area determination unit 210 determines a detection area based on the position of the face part. As described above, the detection areas are, for example, the area from the lower eyelid to the left cheek and the area from the lower eyelid to the right cheek of the right eye.

  In step S1400, the grayscale calculation unit 221 generates a grayscale image from the captured image.

  In step S1500, the smoothing processing unit 222 generates an unsharp image (smoothed image) from the grayscale image.

  Note that the smoothing processing unit 222 may generate the two unsharp images described above.

  In step S1600, the minimum value calculation unit 223 selects a minimum value from the luminance of the pixel of the grayscale image and the luminance of the corresponding pixel of the unsharp image, and generates a high luminance suppression image.

  When two unsharp images are generated in step 1500, the minimum value calculation unit 223 selects a minimum value from the luminances of the corresponding pixels of the two unsharp images and handles the high-intensity suppressed image. It is set as the luminance value of the pixel.

  In step S1700, the difference calculation unit 224 generates a difference image between the high luminance suppression image and the grayscale image, and further synthesizes the difference image and the grayscale image to generate an edge enhanced image.

  When two unsharp images are generated in step 1500, the difference calculation unit 224 calculates a difference between the high-intensity suppressed image and the unsharp image having the higher smoothing strength of the two unsharp images. An image is generated, and the difference image and the grayscale image are combined to generate an edge enhanced image.

  In step S1800, the wrinkle detection processing unit 230 determines a wrinkle region in the base captured image by performing edge detection processing on the edge-enhanced image and comparing the gradient value of each part with a threshold value. Further, the spot detection unit 240 detects a spot area.

  In step S <b> 1900, the image generation unit 150 generates a wrinkle-weighted image and displays the wrinkle-weighted image using the display unit 160.

  By the operation as described above, the eyelid detection apparatus 100 suppresses the influence of the high-luminance portion in the grayscale image, generates an edge enhanced image, detects the eyelid region using the edge enhanced image, and detects the detection result. Can be presented to the user.

<Comparison with conventional unsharp masking>
Here, a difference between wrinkle detection by wrinkle detection apparatus 100 according to the present embodiment and wrinkle detection by conventional unsharp masking will be described with reference to experimental results.

  FIG. 5 is a diagram showing a state of wrinkle detection by conventional unsharp masking in an experiment. Moreover, FIG. 6 is a figure which shows the mode of the wrinkle detection by the wrinkle detection apparatus 100 in experiment.

  Further, as shown in FIGS. 5 and 6, in any of the conventional unsharp masking and wrinkle detection apparatuses 100, a grayscale image 302 and an unsharp image 303 are generated from the captured image 301.

In the conventional unsharp masking, a difference image 311 representing a luminance difference between the grayscale image 302 and the unsharp image 303 is generated. The pixel value D of each part of the difference image 311 is expressed by, for example, the following formula (1) when the luminance value of the grayscale image 302 is represented by the symbol B and the luminance value of the unsharp image 303 is represented by the symbol C.
D = B−C (1)

In the conventional unsharp masking, the grayscale image 302 and the difference image 311 are combined to generate an edge enhanced image 312. The luminance value E of each part of the edge enhanced image 312 is expressed by, for example, the following formula (2). Here, k is a coefficient corresponding to weighting at the time of synthesis.
E = B + kD (2)

  The difference image 311 in the conventional unsharp masking is directly affected by the luminance of the grayscale image 302. For example, in the difference image 311, a region 321 corresponding to a region in the grayscale image 302 where contrast due to skin gloss has occurred has a high value (shown in white in the drawing).

  As a result, strong edges due to skin gloss remain in the edge-enhanced image 312, and in order to prevent erroneous detection due to these, it is necessary to set a high gradient value threshold. However, in this case, detection omission is likely to occur, and the amount of the eyelid region 322 displayed in the eyelid enhancement image 313 is reduced.

On the other hand, wrinkle detection apparatus 100 according to the present embodiment calculates the minimum value between the luminance of grayscale image 302 and the luminance of unsharp image 303 as the pixel value of high-luminance suppression image 310a. The pixel value Xa of each part of the high luminance suppression image 310a is expressed by, for example, the following formula (3).
Xa = min {B, C} (3)

  A high-luminance portion in a region where the high-luminance portion and the low-luminance portion are finely mixed (hereinafter referred to as “contrast region”) is converted into lower luminance by smoothing. Therefore, the pixel value of the unsharp image 303 is used for such a high brightness portion. On the other hand, the low luminance portion in the contrast region is converted to higher luminance by smoothing. Therefore, the pixel value of the gray scale image 302 is used for such a low-luminance portion.

Then, in wrinkle detection apparatus 100, a difference image 311a that represents a difference in luminance between high-luminance suppression image 310a and unsharp image 303 is generated. The pixel value Da of each part of the difference image 311a is expressed by, for example, the following formula (4).
Da = Xa-C (4)

  As described above, in the high luminance suppression image 310a, the pixel value of the unsharp image 303 is used for the high luminance portion, and the pixel value of the gray scale image 302 is used for the low luminance portion. Therefore, the pixel value of each pixel of the difference image 311a is zero in the high luminance portion, and is the difference between the grayscale image 302 and the unsharp image 303 in the low luminance portion. That is, the pixel value of each pixel of the difference image 311a of the present embodiment is the same as that of the conventional difference image 311 in the low luminance part, and the same as that in the case of no edge at all in the high luminance part. .

  Therefore, for example, in the difference image 311a, the pixel value of the region 321a corresponding to the region in which the contrast due to the gloss of the skin has occurred in the grayscale image 302 is larger than that of the conventional difference image 311 (see FIG. 5). , A lower value (shown in white in the figure).

In the wrinkle detection device 100, the grayscale image 302 and the difference image 311a are combined to generate an edge enhanced image 312a. The luminance value Ea of each part of the edge enhanced image 312a is expressed by, for example, the following formula (5).
Ea = B + kDa (5)

  Note that, in the edge enhanced image 312a shown in FIG. 6, the low-luminance portion corresponding to the wrinkles in the grayscale image 302 is more emphasized than in the conventional edge enhanced image 312 shown in FIG. This is because the value of the difference image 311a is emphasized with respect to about 10% of pixels from the largest value. This process is performed in the difference calculation unit 224, for example.

  In the conventional edge-enhanced image 312 shown in FIG. 5, the difference image 311 has a large pixel value corresponding to a region where the contrast due to the gloss of the skin has occurred in the grayscale image 302, so the contrast of the glossy part is enhanced. It has been. On the other hand, in the edge-enhanced image 312a that has been subjected to the above processing, the low-luminance pixels corresponding to the eyelids in the grayscale image 302 are mainly emphasized in the difference image 311a, and the contrast of the eyelid region is enhanced. become.

  As a result, the edge-enhanced image 312 hardly contains strong edges due to skin gloss, and it is not necessary to set a high gradient value threshold. Accordingly, the amount of the haze area 322a displayed in the haze-enhanced image 313a is larger than that of the haze area 322 (see FIG. 5) displayed in the haze-enhanced image 313 by the conventional unsharp masking.

  As described above, it is also confirmed that the wrinkle detection apparatus 100 according to the present embodiment reduces detection omissions while preventing erroneous detection due to gloss compared with conventional unsharp masking. I was able to confirm.

<Effects of the present embodiment>
As described above, wrinkle detection apparatus 100 according to the present embodiment can generate an edge-enhanced image with reduced contrast caused by gloss by suppressing the influence of a portion with high luminance.

  Thereby, the wrinkle detection apparatus 100 can strengthen the contrast generated due to the gloss in the edge-enhanced image relative to the contrast generated due to the wrinkle. Therefore, wrinkle detection apparatus 100 can perform wrinkle detection with high accuracy even when the skin is glossy. That is, it becomes possible to detect wrinkles on the same basis as when a person recognizes the presence or absence of skin wrinkles with the naked eye.

  In addition, wrinkle detection apparatus 100 according to the present embodiment can perform high-precision wrinkle detection even when the skin is glossy. Even if it exists, highly accurate wrinkle detection can be performed.

<Other methods to suppress the influence of high-intensity parts>
Note that the method of generating the edge-enhanced image while suppressing the influence of the high-luminance portion in the image is not limited to the above example.

  For example, the smoothing processing unit 222 may generate an unsharp image by performing smoothing (normalization) at a lower intensity on an image region having a higher skin gloss in the grayscale image. .

  In this case, the difference calculation unit 224 may perform the same process as the conventional unsharp masking using the unsharp image. That is, the difference calculation unit 224 combines the difference between the pixel value of the pixel of the grayscale image and the pixel value of the corresponding pixel of the unsharp image and the pixel value of the pixel of the grayscale image as an edge. The pixel value of the corresponding pixel of the enhanced image is used.

  For example, the difference in luminance between the grayscale image and the unsharp image becomes smaller at the portion smoothed at a lower intensity. Therefore, by the above-described processing, the edge can be weakened in the high luminance region, and the risk of being erroneously detected as a wrinkle can be reduced. Therefore, even with such a configuration, it is possible to detect wrinkles with high accuracy while suppressing the influence of gloss. In this case, wrinkle detection device 100 does not need to include minimum value calculation unit 223.

  When the two unsharp images described above are generated, the smoothing processing unit 222 reduces the intensity of an unsharp image having a low smoothing strength to an image region having a higher skin gloss. The unsharp image may be generated by performing smoothing (normalization). Then, the difference calculation unit 224 combines the difference between the pixel values of the corresponding pixels of the two unsharp images and the pixel value of the pixels of the grayscale image, and the pixel of the corresponding pixel of the edge enhanced image. It may be a value.

  Further, when the smoothing intensity is adaptively changed as described above, the wrinkle detection apparatus 100 acquires information on the gloss intensity distribution (hereinafter referred to as “gloss distribution”) on the skin. It is necessary to have. However, there are various types of information regarding the gloss distribution, how to obtain such information, and how to obtain the gloss distribution from such information.

  The information indicating the gloss distribution includes, for example, the position of the face part, the three-dimensional shape of the skin, the position of the light source with respect to the skin, the region specified by the user, the difference between the image and the orthogonal polarization image of the skin, and the oil amount of the skin Distribution and distribution of the moisture content of the skin.

  For example, the light source that illuminates the skin of the face is usually located above the face. For this reason, the convex face portions such as the cheeks and the forehead tend to be more glossy than the other face portions.

  Therefore, the edge enhancement processing unit 220 determines such a convex face area as an image area with stronger skin gloss, for example, based on the position of the face part. In this case, the edge enhancement processing unit 220 stores in advance an area table that defines the positional relationship between a plurality of feature points of the face and the convex face area, and refers to the area table to Estimate an image area with higher gloss.

  In addition, when the wrinkle detection device 100 includes a three-dimensional shape acquisition unit that acquires a three-dimensional shape of skin from a captured image, the edge enhancement processing unit 220 generates a gloss distribution based on the acquired three-dimensional shape. It may be estimated. Alternatively, the edge enhancement processing unit 220 may store a general three-dimensional model of a face in advance and estimate the gloss distribution based on the model. For example, the edge emphasis processing unit 220 estimates that the level of glossiness is higher than that of other portions for convex portions such as nose tips and cheeks.

  In addition, when the wrinkle detection device 100 includes a light source information acquisition unit that acquires light source information indicating the position of the light source with respect to the skin, the detection region determination unit 210 estimates the gloss distribution based on the acquired light source information. May be. For example, when the light source is on the upper side of the face, the detection area determination unit 210 has a gloss intensity level higher than that of the other parts such as the forehead and the upper part of the cheek where the skin surface faces upward. Estimated high.

  In addition, the difference between the normal image when the skin is illuminated with normal light and the image when the skin is illuminated with polarized light and using an orthogonal polarization filter (orthogonal polarization image) is The higher the gloss, the larger the area.

  Therefore, the gloss information acquisition unit may include, for example, an illumination device and a photographing device that acquire both the normal image and the orthogonal polarization image, and may calculate the difference between the acquired normal image and the orthogonal polarization image. In this case, the edge enhancement processing unit 220 determines a portion where the calculated difference is larger as an image region having a stronger skin gloss.

  In addition, for example, in a portion where the amount of oil and moisture of the skin is higher, the gloss of the skin tends to be strong. Also, usually the forehead and nose have higher skin oil and moisture content than other parts.

  Therefore, the edge enhancement processing unit 220 determines, for example, such a face region having a high skin oil amount or moisture amount as an image region having a higher skin gloss, based on the position of the face part. In this case, the edge enhancement processing unit 220 stores in advance an area table that defines the positional relationship between a plurality of facial feature points and a facial area with a high amount of skin oil or moisture, and refers to the area table. Thus, an image area having a stronger skin gloss is estimated.

  Further, the detection area determination unit 210 may display a captured image on a display with a touch panel, and accept an operation for designating a portion having a higher skin gloss from the user.

  Further, in the wrinkle detection device, the edge enhancement processing unit 220 has a luminance value of a pixel whose luminance value exceeds a predetermined threshold for at least one of a captured image, an unsharp image, a grayscale image, and a difference image. You may perform the process which corrects a value to the value of a threshold value.

  In this case, the difference calculation unit 224 performs the same processing as conventional unsharp masking. That is, the difference calculation unit 224 combines the difference between the pixel value of the pixel of the grayscale image and the pixel value of the corresponding pixel of the unsharp image and the pixel value of the pixel of the grayscale image as an edge. The pixel value of the corresponding pixel of the enhanced image is used. Alternatively, when the two unsharp images described above are generated, the difference calculation unit 224 combines the pixel value difference between the corresponding pixels of the two unsharp images and the pixel value of the pixel of the grayscale image. The value is the pixel value of the corresponding pixel of the edge enhanced image.

  In other words, the above-described processing for correcting the brightness value of the captured image, the unsharp image, the grayscale image, or the difference image is processing for clipping a value that affects the edge-enhanced image in a high brightness region. Therefore, even with such a configuration, it is possible to detect wrinkles with high accuracy while suppressing the influence of gloss. In this case, wrinkle detection device 100 does not need to include minimum value calculation unit 223.

<Other modifications of the present embodiment>
In addition, the eyelid detection device 100 may generate an edge-enhanced image by suppressing the influence of a low-luminance portion in the image in the above-described edge enhancement process in order to detect a region with a high gloss.

  In this case, the edge enhancement processing unit 220 is between the pixel value of the pixel having the maximum luminance among the pixels of the grayscale image and the corresponding pixel of the unsharp image and the pixel value of the pixel of the unsharp image. Calculate the difference. Then, the edge enhancement processing unit 220 sets a value obtained by combining the calculated difference and the pixel value of the pixel of the grayscale image as the pixel value of the corresponding pixel of the edge enhanced image.

In this case, the pixel value Ya of each part of the edge enhanced image is expressed by, for example, the following formula (6).
Ya = max {B, C} (6)

  Further, in the embodiment described above, the detection target is a skin wrinkle, but the detection target may be other scratches such as wall scratches, metal material surface cracks, cloth wrinkles, and the like having a linear groove shape. It may be in a state.

<Summary of this disclosure>
The wrinkle detection device of the present disclosure is a wrinkle detection device that detects a wrinkle region of skin included in an image, and generates an unsharp image by smoothing the acquired image and an image acquisition unit that acquires the image An edge enhancement processing unit that performs an edge enhancement process for generating an edge enhanced image that is an image in which an edge of the image is enhanced based on the generated unsharp image, and the generated edge enhanced image An edge detection processing unit that detects the eyelid region by performing edge detection processing, and the edge enhancement processing suppresses the influence of a portion with high luminance in the image, and It is a process to generate.

  In the wrinkle detection device, the edge enhancement processing includes the pixel value of the pixel having the minimum luminance among the pixels of the image and the corresponding pixel of the unsharp image, and the pixel of the pixel of the unsharp image. And a process of calculating a difference between the value and a value obtained by combining the calculated difference and the pixel value of the pixel of the image as a pixel value of a corresponding pixel of the edge enhanced image. .

  In the wrinkle detection apparatus, the edge enhancement processing generates two unsharp images obtained by smoothing the image with different smoothing intensities, and the luminance is minimized among the corresponding pixels of the two unsharp images. A difference between a pixel value of a pixel to be a value and a pixel value of a corresponding pixel of the unsharp image having the higher smoothing intensity among the two unsharp images, and the calculated You may include the process which makes the value which synthesize | combined the difference and the pixel value of the said pixel of the said image the pixel value of the corresponding pixel of the said edge emphasis image.

  Further, in the wrinkle detection device, the edge enhancement process performs smoothing at a lower intensity on an image region having a higher skin gloss in the image, and the pixel value of the pixel of the image and the image The difference between the pixel value of the corresponding pixel of the unsharp image is calculated, and the value obtained by combining the calculated difference and the pixel value of the pixel of the image is the pixel of the corresponding pixel of the edge enhanced image. It may include processing to be a value.

  In the wrinkle detection apparatus, the edge enhancement processing generates two unsharp images obtained by smoothing the image with different smoothing intensities, and the unsharp having a low smoothing intensities of the two unsharp images. Smoothing is performed at a lower intensity for an image area having a higher gloss of the image, and a difference between the pixel values of corresponding pixels of the two unsharp images is calculated. You may include the process which makes the value which synthesize | combined the pixel value of the said pixel of an image the pixel value of the corresponding pixel of the said edge emphasis image.

  In addition, the wrinkle detection device includes a gloss information acquisition unit that acquires information on a distribution of gloss intensity in the skin, and the edge enhancement processing unit is configured to acquire the information on the skin based on the acquired information. An image area having a higher glossiness is determined, and the image area having a higher skin glossiness is smoothed at a lower intensity in the image, and the information includes the position of the face part, the skin 3 At least one of a three-dimensional shape, a position of a light source with respect to the skin, a region specified by a user, a difference between the image and the orthogonal polarization image of the skin, a distribution of the oil content of the skin, and a distribution of the moisture content of the skin It may be information indicating one.

  In the wrinkle detection apparatus, the edge enhancement processing generates a difference image indicating a difference between the image and the unsharp image, and combines the image and the generated difference image to generate the edge. Including a process of generating an emphasized image, and correcting a luminance value of a pixel whose luminance value exceeds a predetermined threshold value to the threshold value for at least one of the image, the unsharp image, and the difference image May be.

  In the wrinkle detection device, the edge enhancement processing generates two unsharp images obtained by smoothing the image with different smoothing intensities, and generates a difference image indicating a difference between the two unsharp images. Generating the edge-enhanced image by combining the image and the generated difference image, and for at least one of the image, the two unsharp images, and the difference image The luminance value of a pixel whose luminance value exceeds a predetermined threshold value may be corrected to the threshold value.

  In addition, the wrinkle detection device generates a wrinkle image indicating the wrinkle region in the image, and generates an haze-enhanced image in which the generated haze image is superimposed on the image; And a display unit for displaying.

  The wrinkle detection method of the present disclosure is a wrinkle detection method for detecting a wrinkle region of skin included in an image, the step of acquiring the image, and smoothing the acquired image to generate an unsharp image, Based on the generated unsharp image, generating an edge enhanced image that is an image in which an edge of the image is enhanced, and performing edge enhancement processing on the generated edge enhanced image And performing the step of detecting the wrinkle region, and the edge enhancement process is a process of generating the edge enhanced image while suppressing an influence of a portion having a high luminance in the image.

  INDUSTRIAL APPLICABILITY The present invention is useful as a wrinkle detection device and a wrinkle detection method that can perform wrinkle detection with high accuracy even when the skin is glossy.

DESCRIPTION OF SYMBOLS 100 Eyelid detection apparatus 110 Image pick-up part 120 Image acquisition part 130 Face component acquisition part 140 Skin state detection part 150 Image generation part 160 Display part 210 Detection area | region determination part 220 Edge emphasis processing part 221 Gray scale calculation part 222 Smoothing process part 223 Minimum Value calculation unit 224 Difference calculation unit 230 Wrinkle detection processing unit 240 Spot detection unit

Claims (10)

A wrinkle detection device for detecting a wrinkle region of a skin included in an image,
An image acquisition unit for acquiring the image;
Edges subjected to edge enhancement processing, wherein the acquired image is smoothed to generate an unsharp image, and based on the generated unsharp image, an edge-enhanced image that is an image in which the edge of the image is enhanced is generated An emphasis processing unit;
A wrinkle detection processing unit that detects the wrinkle region by performing an edge detection process on the generated edge-enhanced image, and
The edge enhancement process includes:
It is a process of generating the edge-enhanced image by suppressing the influence of a high-luminance part in the image.
皺 Detection device. The edge enhancement process includes:
The difference between the pixel value of the pixel having the minimum luminance among the pixels of the image and the corresponding pixel of the unsharp image and the pixel value of the pixel of the unsharp image is calculated and calculated. Including a process of setting a value obtained by combining the difference and a pixel value of the pixel of the image to a pixel value of a corresponding pixel of the edge-enhanced image,
The wrinkle detection device according to claim 1. The edge enhancement process includes:
Two unsharp images obtained by smoothing the image with different smoothing intensities are generated, a pixel value of a pixel having a minimum luminance among corresponding pixels of the two unsharp images, and the two unsharp images The difference between the pixel value of the corresponding pixel of the unsharp image having the higher smoothing intensity in the image is calculated, and the calculated difference and the pixel value of the pixel of the image are combined. Including a process of setting a value to a pixel value of a corresponding pixel of the edge enhanced image
The wrinkle detection device according to claim 1. The edge enhancement process includes:
Among the images, smoothing is performed at a lower intensity for an image region having a stronger skin gloss, and a pixel value between a pixel value of the pixel of the image and a pixel value of a corresponding pixel of the unsharp image A process of calculating a difference, and combining the calculated difference and the pixel value of the pixel of the image as a pixel value of a corresponding pixel of the edge-enhanced image,
The wrinkle detection device according to claim 1. The edge enhancement process includes:
Two unsharp images obtained by smoothing the image with different smoothing intensities are generated, and an unsharp image having a low smoothing intensity among the two unsharp images has a lower gloss with respect to a stronger image area. Smoothing with intensity, calculating a difference between corresponding pixel values of the two unsharp images, and combining the calculated difference and the pixel value of the pixel of the image, Including a process of setting a pixel value of a corresponding pixel of the edge enhanced image,
The wrinkle detection device according to claim 1. A gloss information acquisition unit that acquires information on the distribution of gloss intensity in the skin;
The edge enhancement processing unit
Based on the acquired information, determine an image area having a stronger skin gloss, smoothing the image area having a stronger skin gloss out of the image at a lower intensity,
The information includes the position of the face part, the three-dimensional shape of the skin, the position of the light source with respect to the skin, the region specified by the user, the difference between the image and the orthogonal polarization image of the skin, and the distribution of the oil content of the skin And information indicating at least one of the moisture distribution of the skin,
The wrinkle detection device according to claim 1. The edge enhancement process includes:
Including a process of generating a difference image indicating a difference between the image and the unsharp image, and generating the edge-enhanced image by combining the image and the generated difference image. For at least one of the sharp image and the difference image, the luminance value of a pixel whose luminance value exceeds a predetermined threshold value is corrected to the threshold value.
The wrinkle detection device according to claim 1. The edge enhancement process includes:
Two unsharp images obtained by smoothing the image with different smoothing intensities are generated, a difference image indicating a difference between the two unsharp images is generated, and the image and the generated difference image are Including a process of combining and generating the edge-enhanced image, and for at least one of the image, the two unsharp images, and the difference image, a luminance value of a pixel whose luminance value exceeds a predetermined threshold value. Correct to the threshold value,
The wrinkle detection device according to claim 1. An image generation unit that generates a haze image indicating the haze region in the image and generates a haze-enhanced image in which the generated haze image is superimposed on the image;
A display unit that displays the wrinkle-enhanced image,
The wrinkle detection device according to claim 1. A wrinkle detection method for detecting a wrinkle region of a skin included in an image,
Obtaining the image;
Smoothing the acquired image to generate an unsharp image, and generating an edge-enhanced image that is an image in which an edge of the image is enhanced based on the generated unsharp image. When,
Detecting the wrinkle region by performing an edge detection process on the generated edge-enhanced image, and
The edge enhancement process includes:
It is a process of generating the edge-enhanced image by suppressing the influence of a high-luminance part in the image.
皺 Detection method.