JP6118214B2 - Wrinkle analysis method and wrinkle analyzer - Google Patents

Description

  The present invention relates to a skin image analysis technique.

  The following Patent Document 1 proposes a method for distinguishing skin properties such as texture composed of skin grooves and cuticles on the skin surface. Patent Document 1 below, as one of the discrimination methods, generates a single color pixel or an image composed of processed pixels obtained by performing arithmetic processing on a single color pixel from a color image indicating a skin state, and Fourier transforms the image. A method is proposed in which an index value obtained from a luminance histogram and / or luminance shape characteristics of a Fourier transform image obtained by applying is used as an index of skin properties.

JP 2006-61170 A

  In the proposed method described above, an index value (standard deviation or the like) obtained from the luminance histogram is used based on the idea that the relative variation characteristic of the hill interval is easily reflected. In other words, the above-described proposed method is a method for discriminating the texture state by quantifying the relative variation characteristics of the skin interval, and does not consider the analysis of the wrinkle state.

  This invention is made | formed in view of such a subject, and provides the technique of quantifying the wrinkle state of skin with high precision based on a skin image.

  Each aspect of the present invention employs the following configurations in order to solve the above-described problems.

  The wrinkle analysis method according to the first aspect acquires a skin luminance image of a site to be analyzed, acquires a binary pattern image representing a local wrinkle texture, and slides the binary pattern image on the skin luminance image while A correlation image is generated by sequentially calculating a correlation value between a value pattern image and a skin luminance image, a binarized image is generated by binarizing the correlation image, and the binarized image is thinned Generating a thin line image and extracting line segment length information from the thin line image.

  The wrinkle analysis device according to the second aspect includes an image acquisition unit that acquires a skin luminance image of a region to be analyzed, a pattern acquisition unit that acquires a binary pattern image representing a local wrinkle texture, and a skin luminance value of the binary pattern image. A correlation processing unit that generates a correlation image by sequentially calculating a correlation value between the binary pattern image and the skin luminance image while sliding on the image, and binarizing the correlation image by binarizing the correlation image Extracted by a binarization unit to be generated, a thinning unit that generates a thinned image by thinning the binarized image, an information extraction unit that extracts line segment length information from the thinned image, and an information extraction unit An output processing unit that performs an output process of the line segment length information.

  As another aspect of the present invention, there may be a program that causes at least one computer to execute the method according to the first aspect, or a computer-readable storage medium that records such a program. Also good. This recording medium includes a non-transitory tangible medium.

  According to each said aspect, the technique which quantifies the wrinkle state of skin with high precision based on a skin image can be provided.

It is a figure which shows notionally the hardware structural example of the wrinkle analyzer in 1st Embodiment. It is a flowchart which shows the operation example of the wrinkle analyzer in 1st Embodiment. It is a figure which shows the example of a binary pattern image. It is a figure which shows the production | generation process of a correlation image. It is a figure which shows the process in which a correlation image is binarized and thinned. It is a figure which shows notion of extraction of line segment length information notionally. It is a figure which shows notionally the process structural example of the wrinkle analyzer in 1st Embodiment. It is a flowchart which shows the operation example of the wrinkle analyzer in 2nd Embodiment. It is a figure which shows notionally the example which calculates the distribution information of the power spectrum for every angle from a skin luminance image. It is a figure which shows the process of angle correction | amendment of a skin luminance image. It is a figure which shows notionally the process structural example of the wrinkle analyzer in 2nd Embodiment. It is a flowchart which shows the operation example of the wrinkle analyzer in 3rd Embodiment. It is a figure which shows the process of a micro area | region removal. It is a figure which shows notionally the process structural example of the wrinkle analyzer in 3rd Embodiment. It is a flowchart which shows the operation example of the wrinkle analyzer in 4th Embodiment. It is a figure which shows notionally the process structural example of the wrinkle analyzer in 4th Embodiment. It is a flowchart which shows the operation example of the wrinkle analyzer in 5th Embodiment. It is a figure which shows notionally the process structural example of the wrinkle analyzer in 5th Embodiment. It is a figure which shows the modification of a binary pattern image. It is a figure which shows the relationship between the image size of a binary pattern image, and a correlation image. It is a graph which shows the correlation with an average line segment pixel number and a wrinkle state. It is a graph which shows correlation with directionality entropy and a wrinkle state.

  Embodiments of the present invention will be described below. In addition, each embodiment given below is an illustration, respectively, and this invention is not limited to the structure of each following embodiment.

[First Embodiment]
[Hardware configuration]
FIG. 1 is a diagram conceptually illustrating a hardware configuration example of a wrinkle analyzer 10 according to the first embodiment. The wrinkle analyzer 10 according to the first embodiment is a so-called computer, and includes a CPU (Central Processing Unit) 2, a memory 3, an input / output interface (I / F) 4, and the like. The memory 3 is a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk, a portable storage medium, or the like.

  The input / output I / F 4 is connected to the input unit 7, the output unit 9, a communication device that communicates with other computers and other devices, and the like. The input unit 7 is a device that receives input of a user operation such as a keyboard and a mouse. The output unit 9 is a device that provides information to a user, such as a display device or a printer. Note that the hardware configuration of the wrinkle analyzer 10 is not limited.

[Operation example (wrinkle analysis method)]
FIG. 2 is a flowchart showing an operation example of the wrinkle analyzer 10 in the first embodiment. Hereinafter, the wrinkle analysis method according to the first embodiment will be described with reference to FIG.
As shown in FIG. 2, the wrinkle analysis method executed by the wrinkle analyzer 10 includes a step of acquiring a skin luminance image (S21), a step of acquiring a binary pattern image (S22), and a step of generating a correlation image. (S23), a binarization step (S24), a thinning step (S25), and an information extraction step (S26). Hereinafter, these steps will be described in detail.

  In the step (S21), the wrinkle analyzer 10 acquires a skin luminance image of the analysis target part. As the analysis target region, for example, a predetermined range under the eye or the corner of the eye where wrinkles that affect the appearance age are likely to occur. However, the analysis target part is not limited. The skin luminance image is an image in which a skin analysis target part is captured, and is an image having light / dark information for each pixel. The skin luminance image may be a gray scale image obtained by converting a color image captured by a general camera into a gray scale. Further, the skin luminance image may be acquired as an image file in a JPEG (Joint Photographic Experts Group) format, a GIF (Graphic Interchange Format) format, or may be acquired as data in which brightness / darkness information of each pixel is listed. . This embodiment does not limit the data format of the skin luminance image. The wrinkle analyzer 10 can acquire the skin luminance image from a portable recording medium, another device, or the like via the input / output I / F 4.

  In the step (S22), the wrinkle analyzer 10 acquires a binary pattern image representing a local wrinkle texture. The wrinkle analyzer 10 may generate a binary pattern image by itself, obtain it from the memory 3 stored in advance, or pass through the input / output I / F 4 from a portable recording medium, another device, or the like. Can also be obtained. The binary pattern image is an image in which the pixel value of each pixel is set to a value corresponding to black or white, and is an image having a linear pattern corresponding to wrinkles as a local wrinkle texture.

  FIG. 3 is a diagram illustrating an example of a binary pattern image. In the example of FIG. 3, the binary pattern image has an image size of 5 pixels vertically and 5 pixels horizontally, and has a linear pattern extending from the upper left to the lower right. The linear pattern is not limited to the example of FIG. For example, in the example of FIG. 3, the linear pattern extends from the upper left to the lower right, but the linear pattern may be in the vertical direction (vertical direction) or in the horizontal direction (horizontal direction). Further, the size of the binary pattern image may be smaller or larger than the example of FIG. Further, the binary pattern image may have a plurality of linear patterns. However, since the linear pattern represented by the binary pattern image corresponds to the wrinkle texture extracted from the skin luminance image, the linear pattern corresponds to the local wrinkle texture, i.e., one wrinkle. It is desirable that the size of the binary pattern image is determined in accordance with the size of the analysis target portion shown in the luminance image, the image size of the skin luminance image, and the like.

  In addition, since the binary pattern image is used to reveal a wrinkle texture (wrinkle component) in the correlation image described later, the direction of the linear pattern that the binary pattern image has and the wrinkle reflected in the skin luminance image It is desirable that the direction matches. Therefore, when a binary pattern image is prepared in advance, the skin luminance image is obtained by imaging the analysis target part so that the direction of wrinkles matches the direction of the linear pattern of the binary pattern image. Is preferably generated.

  In the binary pattern image in the present embodiment, as shown in FIG. 3, a linear pattern is shown in black. This is because the wrinkle texture that appears in the skin brightness image is derived from a dent on the skin surface, and is therefore generally represented by a dark brightness value. Therefore, when the wrinkle texture is represented by a bright luminance value in the skin luminance image, a binary pattern image in which a linear pattern is shown in white can be used. In the following description, a skin luminance image in which the wrinkle texture is expressed by a dark luminance value will be described.

  In the step (S23), the wrinkle analyzer 10 slides the binary pattern image acquired in (S22) on the skin luminance image acquired in (S21), Correlation images are generated by sequentially calculating the correlation values. In generating a correlation image, the wrinkle analyzer 10 linearly assigns the calculated correlation value to pixel values of 0 or more and 255 or less. As a result, each pixel of the generated correlation image has a pixel value corresponding to the calculated correlation value. Thereby, the pixel value obtained from the region in the skin luminance image having a high correlation with the binary pattern image is a relatively bright value, and the pixel value obtained from the region having a low correlation is a relatively dark value. At this time, since the portion showing the correlation value of 0 or less represents a skin texture other than wrinkles, the pixel value of that portion is set to 0. Since the binary pattern image represents a local wrinkle texture, in the correlation image, the wrinkle component is expressed with bright luminance, and other than the wrinkle component is expressed with dark luminance. However, a correlation image may be generated in which the relationship between the correlation value and the pixel value is reversed from the above-described relationship, and the wrinkle component is expressed with dark luminance and the components other than the wrinkle component are expressed with bright luminance.

  FIG. 4 is a diagram showing a correlation image generation process. As shown in FIG. 4, the wrinkle analyzer 10 slides the binary pattern image pixel by pixel on the skin luminance image, and correlates the binary pattern image with the region overlapping the binary pattern image of the skin luminance image. The values are sequentially calculated, and the calculated correlation values are sequentially converted into pixel values. The wrinkle analyzer 10 generates a correlation image by setting a pixel value corresponding to this correlation value for each pixel. The wrinkle analyzer 10 calculates the correlation value using, for example, the following (Expression 1). However, the present embodiment does not limit the correlation value calculation formula to the following formula.

R represents a correlation value. X _i indicates the value of the pixel at position i in the region overlapping the binary pattern image of the skin luminance image. X with an overline indicates an average pixel value in an area overlapping with the binary pattern image of the skin luminance image. Y _i indicates the value of the pixel at position i in the binary pattern image. Y with an overline indicates an average pixel value in the binary pattern image. S indicates the image size (number of pixels) of the binary pattern image.

  In step (S24), the wrinkle analyzer 10 binarizes the correlation image generated in (S23) to generate a binarized image. The wrinkle analyzer 10 determines the binarization threshold using, for example, a discriminant analysis method. The binarization threshold may be determined by other methods such as a mode method, or may be determined in advance and held in the memory 3.

  In step (S25), the wrinkle analyzer 10 thins the binarized image obtained in (S24) to generate a thinned image. Specifically, the wrinkle analyzer 10 determines pixel values having pixel values (white or black) indicating the linear pattern so that the line width of the linear pattern appearing in the binarized image is one pixel. Invert (background). As the thinning algorithm, methods such as Hilditch, Deutsch, Zhang Suen, Tamura, and Tsuruoka may be used.

  FIG. 5 is a diagram illustrating a process in which a correlation image is binarized and thinned. By executing the steps (S24) and (S25) described above, the correlation image is converted into a binarized image, and the binarized image is converted into a thinned image.

  In step (S26), the wrinkle analyzer 10 extracts line segment length information from the thinned image obtained in (S25). The line segment length information is information indicating the characteristics of the length of the line segment appearing in the thinned image. Here, the line segment appearing in the thinned image is a set of line component pixels having pixel values indicating the line segment, and is a pixel set obtained by collecting adjacent line component pixels in at least one of the eight directions. . The length of this line segment can be expressed by the number of line component pixels forming the line segment (hereinafter referred to as the number of line segment pixels), or between two line component pixels located at the end of the line segment. (Hereinafter referred to as a line segment distance).

  Therefore, the wrinkle analyzer 10 can extract the average line segment pixel number, the average line segment distance, the representative line segment number, and the like as the line segment length information from the thinned image. The average number of line segment pixels is the average of the number of line segment pixels in all line segments appearing in the thinned image, and the average line segment distance is the average of the line segment distance in all line segments appearing in the thinned image. The representative line segment number is, for example, the number of line segment pixels or the line segment having the line segment distance larger than a predetermined threshold. In the present embodiment, the line segment length information may be information indicating the characteristics of the length of the line segment appearing in the thinned image, and the specific content of the line segment length information is not limited.

  FIG. 6 is a diagram conceptually illustrating the extraction of line segment length information. FIG. 6 illustrates a state in which the length of a certain line segment is calculated. The wrinkle analyzer 10 detects the line segments (a set of line component pixels) as described above from the thinned image as shown in FIG. 6 and calculates the length of each detected line segment. Based on the length of each line segment, the line segment length information is calculated.

[Processing configuration]
FIG. 7 is a diagram conceptually illustrating a processing configuration example of the wrinkle analyzer 10 according to the first embodiment. The wrinkle analyzer 10 according to the first embodiment includes an image acquisition unit 11, a pattern acquisition unit 12, a correlation processing unit 13, a binarization unit 14, a thinning unit 15, an information extraction unit 16, an output processing unit 17, and the like. Each of these processing units is realized, for example, by executing a program stored in the memory 3 by the CPU 2. Further, the program may be installed from a portable recording medium such as a CD (Compact Disc) or a memory card or another computer on the network via the input / output I / F 4 and stored in the memory 3. Good.

  The wrinkle analyzer 10 can execute the above-described wrinkle analysis method using each processing unit shown in FIG. That is, the wrinkle analyzer 10 can operate as shown in FIG. In this case, the image acquisition unit 11 executes the process (S21), the pattern acquisition unit 12 executes the process (S22), the correlation processing unit 13 executes the process (S23), and the binarization unit 14 performs the process (S22). Step S24) is executed, the thinning unit 15 executes the step (S25), and the information extraction unit 16 executes the step (S26).

  The output processing unit 17 performs output processing of the line segment length information extracted by the information extraction unit 16. For example, the output processing unit 17 causes the output unit 9 to output line segment length information. This output includes screen display, printing, audio output, and the like. Further, the output processing unit 17 may store (output) the data of the line segment length information in the portable recording medium via the input / output I / F 4 or transmit (output) the data to other devices. Good. Thus, this embodiment does not restrict the output form of line segment length information.

  The wrinkle analyzer 10 may hold the skin luminance image of the analysis target part in advance in the wrinkle analyzer 10 itself, may be generated by the wrinkle analyzer 10 itself, or may be another computer or portable recording. You may acquire from media etc. via input / output I / F4.

[Operation and Effect in First Embodiment]
As described above, in the first embodiment, a correlation image having a pixel value corresponding to a correlation value between a skin luminance image in which a region to be analyzed is captured and a binary pattern image representing a local wrinkle texture is generated. Is binarized and thinned. Then, line segment length information is extracted from the thinned image. As the extracted line segment length information, line segment length information, which is information indicating the characteristics of the length of the line segment appearing in the thinned image, is extracted.

  Since the binary pattern image represents a local wrinkle texture, the wrinkle texture included in the analysis target portion is revealed by the correlation image, and the manifested wrinkle texture is obtained by binarization and thinning of the correlation image. It is represented as a line segment in the thinned image. Therefore, the line segment length information extracted from the thinned image is information indicating the characteristics of the length of wrinkles included in the analysis target part.

  The present inventors considered that the wrinkle length is one index of the wrinkle state, and found that the wrinkle state can be quantified by the line segment length information obtained by the image analysis as described above. Then, the inventors verified that there is a correlation between the line segment length information and the wrinkle index value obtained by quantifying the wrinkle state, as will be described later as an example. Here, the wrinkle state means the state of the texture of the skin that is visually recognized when there are many or few wrinkles, and can also be expressed as a wrinkle feeling.

  Therefore, according to the first embodiment, one wrinkle index value (line segment length information) whose wrinkle state is quantified is extracted by simple image processing without using a complicated physical model or special equipment. can do.

[Second Embodiment]
Hereinafter, the wrinkle analysis device 10 and the wrinkle analysis method according to the second embodiment will be described focusing on the content different from the first embodiment. In the second embodiment, a process for correcting the angle of the skin luminance image is newly added. Hereinafter, the description of the same content as the first embodiment will be omitted as appropriate. The hardware configuration of the wrinkle analyzer 10 in the second embodiment is the same as that of the first embodiment shown in FIG.

[Operation example (wrinkle analysis method)]
Hereinafter, the wrinkle analysis method according to the second embodiment will be described with reference to FIG. FIG. 8 is a flowchart illustrating an operation example of the wrinkle analyzer 10 according to the second embodiment. In FIG. 8, the same steps as those in FIG.
The wrinkle analysis method in the second embodiment executed by the wrinkle analyzer 10 further includes steps (S51) and (S52) in addition to the method of the first embodiment shown in FIG.

  In the step (S51), the wrinkle analyzer 10 acquires angle information related to the directionality of wrinkles shown in the skin luminance image (S51). Angle information related to the direction of wrinkles in the skin luminance image may be input by the user operating the input unit 7 based on an input screen or the like. For example, the wrinkle analyzer 10 can output an input screen that allows the user to input an angle value. In addition, the wrinkle analyzer 10 displays the original skin luminance image on the output unit 9, detects an operation of rotating the skin luminance image by the user, and acquires the angle information from the detected operation. it can. However, the method for obtaining angle information regarding the directionality of wrinkles in the skin luminance image is not limited.

  The wrinkle analyzer 10 can also calculate power spectrum distribution information for each angle by applying Fourier transform to the skin luminance image, and obtain the angle information from the distribution information. As the Fourier transform applied to the skin luminance image, for example, Fast Fourier Transform (FFT) and Discrete Fourier Transform (DFT) are well known. The wrinkle analyzer 10 calculates a power spectrum by applying a Fourier transform (two-dimensional Fourier transform) to the vertical direction and the horizontal direction of the skin luminance image, and calculates power spectrum distribution information for each angle from the power spectrum. Extract. The extracted distribution information may be set in a range greater than 0 degrees and less than 180 degrees. It is only necessary to detect the direction of wrinkles in the skin luminance image, and information obtained from an angle range greater than 180 degrees and less than 360 degrees can be considered equivalent to information in an angle range greater than 0 degrees and less than 180 degrees. Because.

  For example, the wrinkle analyzer 10 may acquire an angle indicating a peak as the angle information from the distribution information excluding information in the 0 degree direction and the 90 degree direction in the distribution information. The components of the 0 degree direction (including the 180 degree direction) and the 90 degree direction (including the 270 degree direction) correspond to the vertical and horizontal components of the image, and are wrinkled due to the characteristics of Fourier transform that calculates the image as a periodic signal. Regardless, since it shows a high value, there is no problem even if those angular components are excluded.

  FIG. 9 is a diagram conceptually illustrating an example of calculating power spectrum distribution information for each angle from the skin luminance image. In FIG. 9, an angle histogram is shown as distribution information. In the example of FIG. 9, the angular interval of the distribution information is set to 1 degree, and the distribution information of the power spectrum for each angle from 1 degree to 179 degrees excluding 90 degrees is calculated. According to the distribution information (angle histogram) shown in the example of FIG. 9, since the peak of the power spectrum occurs around 125 degrees, for example, angle information indicating the directionality of wrinkles reflected in the skin luminance image is 125 degrees. Get as.

  In step (S52), the wrinkle analyzer 10 corrects the angle of the skin luminance image acquired in (S21) based on the angle information acquired in (S51). In other words, the wrinkle analysis device 10 corrects the angle of the skin luminance image based on the directionality of the wrinkles that appear in the skin luminance image. Specifically, the wrinkle analyzer 10 is configured so that the wrinkle direction indicated by the angle information acquired in (S51) matches the direction of the local wrinkle texture (linear pattern) represented by the binary pattern image. In addition, the angle of the skin luminance image is corrected.

  FIG. 10 is a diagram illustrating a process of angle correction of a skin luminance image. As shown in FIG. 10, the wrinkle analyzer 10 rotates the skin luminance image so that the direction of wrinkles in the skin luminance image faces a predetermined angle direction. In the example of FIG. 9, the predetermined angle direction is set to a diagonal 45 degree direction connecting the upper left and the lower right, and the wrinkle directionality indicates the direction of about 30 degrees oblique connecting the upper left and the lower right. The skin luminance image is rotated about 15 degrees clockwise. In the example of FIG. 9, a non-rotated square area is cut out from the rotated image, and the area is used as a normalized image in subsequent processing.

  In step (S23), the wrinkle analyzer 10 generates a correlation image from the skin luminance image whose angle has been corrected in (S52), as in the first embodiment.

[Processing configuration]
FIG. 11 is a diagram conceptually illustrating a processing configuration example of the wrinkle analyzer 10 in the second embodiment. The wrinkle analyzer 10 according to the second embodiment further includes an angle information acquisition unit 21 and an angle correction unit 22 in addition to the configuration of the first embodiment. The angle information acquisition unit 21 and the angle correction unit 22 are also realized by executing a program stored in the memory 3 by the CPU 2, similarly to the other processing units.

  The wrinkle analyzer 10 can execute the above-described wrinkle analysis method using each processing unit shown in FIG. That is, the wrinkle analyzer 10 can operate as shown in FIG. In this case, the angle information acquisition unit 21 executes the step (S51), and the angle correction unit 22 executes the step (S52). Moreover, the angle information acquisition part 21 can also calculate the distribution information of the power spectrum for every angle by applying a Fourier transformation with respect to a skin luminance image as a process (S51). The correlation processing unit 13 generates a correlation image using the skin luminance image whose angle is corrected by the angle correction unit 22 and the binary pattern image acquired by the pattern acquisition unit 12. Other processing units are the same as those in the first embodiment.

[Operations and effects in the second embodiment]
As described above, in the second embodiment, the skin luminance image is subjected to angle correction based on the directionality of the wrinkles appearing in the skin luminance image, and a correlation image is generated from the angle corrected skin luminance image. Thus, according to the second embodiment, even if the directionality of the wrinkle reflected in the skin luminance image does not match the directionality of the local wrinkle texture of the binary pattern image prepared in advance, each directionality is Since the skin luminance image is angle-corrected so as to match, it is possible to eliminate the limitation of the imaging method of the skin luminance image. That is, even if a skin luminance image is generated by imaging an analysis target part without considering wrinkle directionality, according to the second embodiment, the line segment length that becomes one index value of the wrinkle state Information can be extracted with high accuracy.

  Furthermore, the angle information indicating the directionality of wrinkles in the skin luminance image can also be obtained by performing image analysis on the skin luminance image. Specifically, power spectrum distribution information for each angle is calculated by applying Fourier transform to the skin luminance image, and the angle information can be acquired from this distribution information. By using image analysis in this manner, angle information used for angle correction of a skin luminance image can be acquired with high accuracy, and line segment length information can be extracted with high accuracy.

[Third Embodiment]
Hereinafter, the wrinkle analysis apparatus 10 and the wrinkle analysis method according to the third embodiment will be described focusing on the content different from the above-described embodiments. In the third embodiment, a skin luminance image smoothing process, a difference image calculation process, and a minute area pixel removal process are newly added. In the following, a mode in which a skin luminance image smoothing process, a difference image calculation process, and a minute area pixel removal process are added to the above-described second embodiment will be described as a third embodiment. However, these new processes may be added to the first embodiment. Hereinafter, descriptions of the same contents as those in the first embodiment and the second embodiment are omitted as appropriate. The hardware configuration of the wrinkle analyzer 10 in the third embodiment is the same as that of the first embodiment shown in FIG.

[Operation example (wrinkle analysis method)]
Hereinafter, the wrinkle analysis method according to the third embodiment will be described with reference to FIG. FIG. 12 is a flowchart illustrating an operation example of the wrinkle analyzer 10 according to the third embodiment.

  The wrinkle analysis method in the third embodiment executed by the wrinkle analyzer 10 further includes a step (S61), a step (S62), and a step (S63) in addition to the second embodiment. However, FIG. 12 shows an example in which new steps (S61), (S62), and (S63) in the third embodiment are added to the method of the second embodiment, but the new steps (S61), (S62) are shown. ) And (S63) can also be added to the method of the first embodiment shown in FIG. In FIG. 12, the same steps as those in FIGS. 2 and 8 are denoted by the same reference numerals.

  In step (S61), the wrinkle analyzer 10 applies a smoothing filter to the skin luminance image acquired in (S21). The smoothing filter is a filter that smoothes the distribution of luminance values of adjacent pixels, and there are a moving average filter, a weighted average filter, a Gaussian filter, and the like. In the present embodiment, for example, a Gaussian filter is used.

  In step (S62), the wrinkle analyzer 10 calculates a difference image between the skin luminance image smoothed in (S61) and the skin luminance image acquired in (S21). Thereby, in the process (S52), the wrinkle analyzer 10 performs angle correction on the difference image calculated in (S62). However, (S61) and (S62) may be performed after the step (S52). In this case, the wrinkle analyzer 10 applies a smoothing filter to the skin luminance image whose angle has been corrected.

  In step (S63), the wrinkle analyzer 10 inverts the pixel value of the minute region pixel in the pixel having the pixel value representing the wrinkle texture in the binarized image generated in (S24). Here, the minute area pixel means a pixel included in a small area (small area) where a line segment cannot be formed, which is an area formed by one or more pixels having a pixel value representing a wrinkle texture. To do. Since such a small area is likely to be texture or noise other than wrinkles, the wrinkle analyzer 10 inverts the pixel values of the pixels included in such a small area to remove the small area. To do.

  For example, the wrinkle analyzer 10 can detect a minute region pixel from the wrinkle component pixels having pixel values representing a wrinkle texture as follows. The wrinkle analyzer 10 detects wrinkle component pixels that are not adjacent to other wrinkle component pixels in any of the eight directions as minute area pixels. In addition, the wrinkle analyzer 10 detects a set in which the number of pixels included in a set of wrinkle component pixels adjacent in at least one of eight directions is a predetermined number (for example, 3) or less as a minute region, and is included in the set The wrinkle component pixels to be detected are detected as minute region pixels.

In step (S25), the wrinkle analyzer 10 thins the binarized image obtained by inverting the pixel values of the minute region pixels in (S63).
FIG. 13 is a diagram illustrating a process of removing a minute region. As illustrated in FIG. 13, the binarized image includes a minute region (white region) where a line segment cannot be formed. In the binarized image from which the minute area is removed, the minute area is blackened by inverting the pixel value of the minute area pixel. The binarized image from which the minute area has been removed is thinned to generate a thinned image.

[Processing configuration]
FIG. 14 is a diagram conceptually illustrating a processing configuration example of the wrinkle analyzer 10 according to the third embodiment. The wrinkle analysis device 10 according to the third embodiment further includes a smoothing processing unit 24, a difference calculation unit 25, and a minute region removal unit 26 in addition to the configuration of the second embodiment. As with the other processing units, the smoothing processing unit 24, the difference calculation unit 25, and the minute region removal unit 26 are also realized by executing a program stored in the memory 3 by the CPU 2.

  The wrinkle analyzer 10 can execute the above-described wrinkle analysis method using each processing unit shown in FIG. That is, the wrinkle analyzer 10 can operate as shown in FIG. In this case, the smoothing processing unit 24 executes the step (S61), the difference calculating unit 25 executes the step (S62), and the minute region removing unit 26 executes the step (S63). The angle correction unit 22 corrects the angle of the difference image calculated by the difference calculation unit 25. The thinning unit 15 thins the binarized image in which the minute region pixels are inverted by the minute region removing unit 26. Other processing units are the same as those in the second embodiment.

[Operations and effects in the third embodiment]
In the third embodiment, a smoothing filter is applied to the skin luminance image, and a difference image between the smoothed skin luminance image and the original skin luminance image is calculated. By applying the smoothing filter, it is possible to extract a skin texture (noise texture) derived from other than the uneven shape of the skin surface from the skin luminance image. The extracted noise texture includes, for example, color unevenness components such as spots, freckles, and moles, and light source unevenness components. Then, by calculating the difference image, such a noise texture can be removed from the original skin luminance image, and as a result, the skin texture derived from the uneven shape of the skin surface is extracted with high purity as the difference image. Can do. In the third embodiment, a correlation image is generated from this difference image.

  Therefore, according to the third embodiment, since the line segment length information is extracted from the image from which the skin texture that is not related to the wrinkle component is removed, the wrinkle analysis accuracy can be improved.

  Furthermore, in the third embodiment, the pixel values of the minute region pixels are inverted in the binarized image, so that the minute region that is likely to represent texture or noise other than wrinkles is removed from the binarized image. Is done. Therefore, according to the third embodiment, since the line segment length information is extracted from the image from which the skin texture that is not related to the wrinkle component is removed, the wrinkle analysis accuracy can be improved.

[Fourth Embodiment]
Hereinafter, the wrinkle analysis apparatus 10 and the wrinkle analysis method according to the fourth embodiment will be described focusing on the contents different from the above-described embodiments. The fourth embodiment is different from the above-described embodiments in that a wrinkle index value corresponding to line segment length information is calculated. Hereinafter, the description of the same contents as the above-described embodiments will be omitted as appropriate. The hardware configuration of the wrinkle analyzer 10 in the fourth embodiment is the same as that of the first embodiment shown in FIG.

[Operation example (wrinkle analysis method)]
Hereinafter, the wrinkle analysis method according to the third embodiment will be described with reference to FIG.
FIG. 15 is a flowchart illustrating an operation example of the wrinkle analyzer 10 according to the fourth embodiment.

  The wrinkle analysis method according to the fourth embodiment executed by the wrinkle analyzer 10 further includes a step (S71) in addition to the above-described embodiments. However, FIG. 15 shows an example in which a new step (S71) in the fourth embodiment is added to the method of the third embodiment, but the step (S71) is the method of the first embodiment shown in FIG. And it can be added to the method of the second embodiment shown in FIG. In FIG. 15, the same steps as those in FIG. Hereinafter, each process will be described in detail with a focus on processes having different contents from the above-described embodiments.

  In the step (S71), the wrinkle analyzer 10 generates wrinkles corresponding to the line length information extracted in (S26) based on the correlation between the line length information and the wrinkle index value in which the wrinkle state is quantified. An index value is calculated. For example, when the average line segment pixel count is calculated as the line segment length information in (S26), the wrinkle analyzer 10 correlates the average line segment pixel count and the wrinkle index value in which the wrinkle state is quantified. Based on the above, the wrinkle index value corresponding to the average number of line segment pixels extracted in (S26) is calculated. For example, the wrinkle index value is set to a value that becomes larger as the wrinkle state is visually recognized when there are many wrinkles, and becomes smaller as the wrinkle state that is visually recognized when there are few wrinkles. It is desirable. The correlation between the line segment length information and the wrinkle index value is converted into data in a predetermined format such as a function format or a table format and stored in the memory 3. Details of this correlation will be described in detail in the Examples section.

[Processing configuration]
FIG. 16 is a diagram conceptually illustrating a processing configuration example of the wrinkle analyzer 10 according to the fourth embodiment. The wrinkle analyzer 10 according to the fourth embodiment further includes an index value calculation unit 28 in addition to the configurations of the above-described embodiments. The index value calculation unit 28 is also realized by executing a program stored in the memory 3 by the CPU 2 in the same manner as other processing units. However, FIG. 16 illustrates an example of a processing configuration in which an index value calculation unit 28 is newly added to the processing configuration of the third embodiment.

  The wrinkle analyzer 10 can execute the above-described wrinkle analysis method using each processing unit shown in FIG. The wrinkle analyzer 10 can operate as shown in FIG. In this case, the index value calculation unit 28 executes the above-described step (S71). The output processing unit 17 performs output processing of at least one of the line segment length information and the wrinkle index value calculated by the index value calculation unit 28. Other processing units are the same as those in the above-described embodiments.

[Operations and effects in the fourth embodiment]
In the fourth embodiment, a wrinkle index value converted from line segment length information is calculated. Therefore, according to the fourth embodiment, it is possible to calculate and output a quantitative value of the wrinkle state that is easier for the person to be evaluated to grasp than the line segment length information.

[Fifth Embodiment]
Hereinafter, the wrinkle analysis device 10 and the wrinkle analysis method according to the fifth embodiment will be described focusing on the content different from the above-described embodiments. The fifth embodiment is different from the above-described embodiments in that the directional entropy is calculated as another index indicating the wrinkle state, and the wrinkle state is comprehensively determined from the directional entropy and the line segment length information. . Hereinafter, the description of the same contents as the above-described embodiments will be omitted as appropriate. The hardware configuration of the wrinkle analyzer 10 in the fifth embodiment is the same as that of the first embodiment shown in FIG.

[Operation example (wrinkle analysis method)]
Hereinafter, the wrinkle analysis method according to the fifth embodiment will be described with reference to FIG.
FIG. 17 is a flowchart illustrating an operation example of the wrinkle analyzer 10 according to the fifth embodiment.

  The wrinkle analysis method according to the fifth embodiment executed by the wrinkle analyzer 10 further includes steps (S81) and (S82) in addition to the fourth embodiment. However, FIG. 17 shows an example in which new steps (S81) and (S82) in the fifth embodiment are added to the method of the fourth embodiment. However, the new steps are the same as those in the first embodiment to the third embodiment. It can also be added to each method of the embodiment. In FIG. 17, the same steps as those in FIG. 15 are denoted by the same reference numerals.

  In step (S81), the wrinkle analyzer 10 calculates power spectrum distribution information for each angle by applying Fourier transform to the difference image calculated in (S62). The calculation of this distribution information is as described above. However, the angular interval of the distribution information is not limited. The smaller the angular interval, the better the analysis accuracy. Therefore, it is desirable to set the angular interval to a small interval of less than 5 degrees.

  In the step (S81), it is desirable that the wrinkle analyzer 10 calculates the distribution information in a state where components in the 0 degree direction and the 90 degree direction are removed. This is because, as described above, the components in the 0 degree direction (including the 180 degree direction) and the 90 degree direction (including the 270 degree direction) show high values regardless of the wrinkle state due to the characteristics of the Fourier transform. is there. By excluding the vertical and horizontal components in this way, noise components other than the wrinkle component can be removed from the distribution information, and the accuracy of the wrinkle analysis can be improved. The wrinkle analyzer 10 may calculate distribution information in a range of 0 degrees or more and less than 180 degrees, and then remove components of 0 degrees and 90 degrees from the distribution information. In addition, the angular range of the distribution information may be set to a range greater than 0 degree and less than 180 degrees. It is only necessary to detect the directionality of the wrinkle texture reflected in the skin luminance image, and a component obtained from an angle range greater than 180 degrees and less than 360 degrees may be considered equivalent to a component in an angle range greater than 0 degrees and less than 180 degrees. Because it can.

In step (S82), the wrinkle analyzer 10 calculates directional entropy indicating the degree of bias of the distribution of the power spectrum for each angle from the distribution information calculated in (S81). For example, the power spectrum distribution for each angle calculated in (S81) is assumed to be a probability density distribution, and the directional entropy is calculated as the information entropy H of the random variable (angle) by the following equation. In the following formula, p _i is a histogram value of an angle i, n is the number of data, and is set to 180. The directional entropy calculated by the above formula takes a small value when the histogram is biased to a specific angle, and takes a large value when the histogram is totally dispersed. However, the calculation method of the directional entropy is not limited to the following formula as long as it is a method that can calculate the magnitude of the bias of the distribution of the power spectrum for each angle.

  In step (S71), the wrinkle analyzer 10 uses the line segment length information extracted in (S26) and the directional entropy calculated in (S82) to quantify the wrinkle index value. Is calculated (S83). At this time, the wrinkle analyzer 10 weights each of the line segment length information and the directional entropy, and calculates a wrinkle index value based on a statistical value (sum, product, average, etc.) of each weighted value. can do.

[Processing configuration]
FIG. 18 is a diagram conceptually illustrating a processing configuration example of the wrinkle analyzer 10 according to the fifth embodiment. The wrinkle analyzer 10 according to the fifth embodiment further includes a distribution calculation unit 31 and an entropy calculation unit 32 in addition to the configurations of the above-described embodiments. Similarly to the other processing units, the distribution calculation unit 31 and the entropy calculation unit 32 are realized by executing a program stored in the memory 3 by the CPU 2. However, FIG. 18 illustrates an example of a processing configuration in which a distribution calculation unit 31 and an entropy calculation unit 32 are newly added to the processing configuration of the fourth embodiment.

  The wrinkle analyzer 10 can execute the above-described wrinkle analysis method using each processing unit shown in FIG. The wrinkle analyzer 10 can operate as shown in FIG. In this case, the distribution calculation unit 31 executes the above-described step (S81), and the entropy calculation unit 32 executes the above-described step (S82). Further, the index value calculation unit 28 calculates a wrinkle index value in which the wrinkle state is quantified based on the line segment length information extracted by the information extraction unit 16 and the directional entropy calculated by the entropy calculation unit 32. calculate. Other processing units are the same as those in the above-described embodiments.

[Operations and effects in the fifth embodiment]
As described above, in the fifth embodiment, the Fourier transform is applied to the skin luminance image in which the site to be analyzed is captured, thereby calculating the power spectrum distribution information for each angle, and the directional entropy of this distribution information. Is calculated. The distribution of the power spectrum for each angle indicates the direction of the texture of the skin (analysis target part) that appears in the skin luminance image. When a texture having a strong directionality exists, a high frequency component appears in a specific direction orthogonal to the texture direction. On the other hand, when the skin texture has no directionality, the high frequency components are not biased in a specific direction, and the high frequency components are dispersed. Therefore, it can be said that the directional entropy indicating the degree of bias of the distribution of the power spectrum for each angle indicates the strength of the directionality of the skin texture reflected in the skin luminance image.

  The present inventors considered that wrinkles are directional skin textures, and further found that wrinkle states can be quantified also by directional entropy obtained by image analysis as described above. That is, the present inventors can separately present line segment length information, which is a feature amount of wrinkle length, and directional entropy, which is a feature amount of wrinkle direction, as an index representing the wrinkle state. I found.

  In the fifth embodiment, a wrinkle index value in which the wrinkle state is quantified is calculated from the line segment length information and the directional entropy. Therefore, according to the fifth embodiment, since the total wrinkle index value is calculated based on the two feature amounts of the wrinkle length and the wrinkle directionality, the index value reflecting the wrinkle state with high accuracy is obtained. Can be calculated.

[Modification]
In each of the embodiments described above, the wrinkle analyzer 10 may acquire a predetermined binary pattern image in a fixed manner, or obtain a binary pattern image suitable for a skin luminance image from a plurality of binary pattern images. You may make it select. Specifically, the wrinkle analyzer 10 corresponds to the directionality of wrinkles that appear in the skin luminance image from a plurality of binary pattern images that respectively represent local wrinkle textures extending in a predetermined angle direction for each of a plurality of predetermined angles. A binary pattern image may be selected.

  FIG. 19 is a diagram illustrating a modification of the binary pattern image. In the above case, in addition to the binary pattern image corresponding to the diagonal direction from the upper left to the lower right shown in FIG. 3, as shown in FIG. 19, a plurality of corresponding to the vertical direction, the horizontal direction, the reverse diagonal direction, etc. Binary pattern images are provided. The wrinkle analyzer 10 may generate such a plurality of binary pattern images by itself, obtain it from the memory 3 held in advance, or input / output I from a portable recording medium, another device, or the like. It can also be acquired via / F4.

  Furthermore, the wrinkle analyzer 10 estimates the directionality (angle) of the wrinkle in the skin luminance image using the distribution information of the power spectrum for each angle obtained from the skin luminance image as in the second embodiment, and A binary pattern image corresponding to this angle can be selected from the plurality of binary pattern images. In this case, the wrinkle analyzer 10 may acquire an angle indicating a peak as the angle information from the distribution information excluding information in the 0 degree direction and the 90 degree direction in the distribution information.

[Supplement]
In the plurality of flowcharts used in the above description, a plurality of steps (processes) are described in order, but the execution order of the steps executed in each embodiment is not limited to the description order. The order of the steps shown in the figure can be changed within a range that does not hinder the contents.

  The impression determination method of each embodiment described above may include a process performed by a person, and transition between processes may be performed by a person's judgment or operation. For example, the wrinkle analyzer 10 outputs a thin line image as it is as line segment length information, and from the output thin line image, a human character of the length of the line segment appearing in the thin line image (line segment length information) ) May be visually recognized. Moreover, selection of one binary pattern image corresponding to a skin luminance image from a plurality of binary pattern images may be performed by a person. In this case, the practitioner can visually recognize the directionality of the wrinkles in the skin luminance image and can select an appropriate binary pattern image according to the directionality. Furthermore, the removal of the minute area pixels may be performed by a person. In this case, the practitioner looks at the binarized image, finds a minute area from the binarized image, and designates pixels that form the minute area. The wrinkle analyzer 10 may present the binarized image to the practitioner and invert the pixel value of the micro area pixel specified by the practitioner.

  The wrinkle analysis method described above uses the correlation between objective data (line segment length information) obtained by image processing on a skin luminance image and wrinkle state statistical data received by a person from the skin luminance image. Thus, it is a method of repeatedly and continuously realizing a certain effect that the wrinkle state can be easily analyzed only by image processing. Therefore, even if a process carried out by a person is included, the wrinkle analysis method described above can be said to be a technical idea utilizing the laws of nature as a whole.

  Moreover, each above-mentioned embodiment and each modification can be combined in the range with which the content does not conflict.

  Examples will be given below, and the above-described embodiments and modifications will be described more specifically. The present invention is not limited in any way by the following examples.

  FIG. 20 is a diagram illustrating the relationship between the image size of the binary pattern image and the correlation image. Reference signs P1, P2, and P3 indicate correlation images, and reference signs PT1, PT2, and PT3 indicate binary pattern images. The binary pattern image PT1 has an image size of 3 × 3, the binary pattern image PT2 has an image size of 5 × 5, and the binary pattern image PT3 has an image size of 7 × 7. Correlation images P1, P2, and P3 are correlations between a skin luminance image having an image size of 64 × 64, generated using the above (Equation 1), and binary pattern images PT1, PT2, and PT3 shown below. It is an image.

  According to the example of FIG. 20, many wrinkle textures are extracted regardless of which binary pattern image is used. In other words, many components other than the wrinkle texture have been extracted from the correlation image P1, and a wrinkle texture that cannot be extracted from the correlation image P3 can be seen. Therefore, according to the example of FIG. 20, it is desirable that the binary pattern image PT2 having an image size of 5 × 5 is used for the analysis. As described above, it is desirable that the image size of the binary pattern image is set to an appropriate size so as to extract many wrinkle textures with less noise components according to the acquired skin luminance image.

  Furthermore, by sensory evaluation using skin luminance images of 20 persons, there is a correlation between the average number of line segment pixels as the line segment length information and the wrinkle state, and a correlation between the directional entropy and the wrinkle state. Has been verified to exist. In this verification, a predetermined region under the eyes is selected as a region to be analyzed, 20 skin luminance images are shown to 3 evaluators, and each evaluator is presented in ascending order from an image with a small amount of wrinkles visually recognized to an image with a large amount of wrinkles. Ranking was made, and the average ranking was set as the wrinkle state point of each skin luminance image. On the other hand, the average line segment pixel number (line segment length information) and directional entropy of each skin luminance image were calculated by the method of the fifth embodiment described above.

  FIG. 21 is a graph showing the correlation between the average number of line segment pixels and the wrinkle state. FIG. 22 is a graph showing the correlation between directional entropy and wrinkle state. As a result of the sensory evaluation, the horizontal axis represents the wrinkle state point, the vertical axis represents the average line segment pixel number, the horizontal axis represents the wrinkle state point, and the vertical axis represents the directional entropy. On the two-dimensional graph, points (diamond dots) corresponding to 20 skin luminance images were plotted. As a result, as shown in FIGS. 21 and 22, there is a positive correlation (correlation coefficient r = 0.574) between the average number of line segment pixels and the wrinkle state, and the directional entropy and the wrinkle state It was proved that there was a negative correlation (correlation coefficient r = −0.565). It was verified that both the line segment length information and the directional entropy calculated as described above can be wrinkle index values representing the wrinkle state.

  In the fourth embodiment described above, the wrinkle index value corresponding to the line segment length information is calculated using the correlation shown in FIG. 21 or the correlation obtained by the method shown in FIG. Good. In the example of FIG. 21, the wrinkle state point may be calculated as the wrinkle index value. In the present embodiment, only the example in which the average line segment pixel number is used as the line segment length information is shown, but even if the average line segment distance, the representative line segment number, etc. are used as the line segment length information, Correlates with the wrinkle state.

  Moreover, the skin luminance image used in the above-mentioned sensory evaluation was imaged in a state in which illumination was illuminated from a direction of 40 degrees from the normal direction of the analysis target site of the subject's skin. In each of the above-described embodiments, modifications, and examples, since the skin texture (wrinkle state) derived from the shape of the skin surface is an analysis target, the illumination angle is set so that the analysis target skin texture is easy to appear in the image. Is preferably adjusted. For example, it is desirable that the illumination angle is set in a range of 30 degrees to 50 degrees from the normal direction of the skin analysis target region.

  A part or all of the above-described embodiments, modifications, and examples may be specified as follows. However, each above-mentioned embodiment, a modification, and an Example are not restrict | limited to the following description.

<1> Obtain a skin luminance image of the site to be analyzed,
Obtain a binary pattern image representing the local wrinkle texture,
While calculating the correlation value between the binary pattern image and the skin luminance image while sliding the binary pattern image on the skin luminance image, a correlation image is generated,
A binarized image is generated by binarizing the correlation image,
A thinned image is generated by thinning the binarized image,
Extracting line segment length information from the thinned image;
Wrinkle analysis method.

<2> Based on the direction of wrinkles in the skin luminance image, the skin luminance image is angle-corrected.
Further including
The wrinkle analysis method according to <1>, wherein the correlation image is generated from the angle-corrected skin luminance image.
<3> By applying Fourier transform to the skin luminance image, power spectrum distribution information for each angle is calculated.
Further including
In the angle correction, the skin luminance image is angle-corrected so that the directionality of wrinkles obtained from the distribution information matches the directionality of the local wrinkle texture represented by the binary pattern image. <2> The wrinkle analysis method described.
<4> The acquisition of the binary pattern image is the directionality of wrinkles reflected in the skin luminance image from a plurality of binary pattern images each representing a local wrinkle texture extending in the predetermined angle direction for each of a plurality of predetermined angles. The wrinkle analysis method according to <1>, wherein a binary pattern image corresponding to is selected.
<5> By applying Fourier transform to the skin luminance image, power spectrum distribution information for each angle is calculated.
Further including
The wrinkle analysis method according to <4>, wherein the acquisition of the binary pattern image selects a binary pattern image corresponding to the distribution information from the plurality of binary pattern images.
<6> In the binarized image, the pixel value of the minute region pixel in the pixel having the pixel value representing the wrinkle texture is inverted.
Further including
The wrinkle analysis method according to any one of <1> to <5>, wherein the generation of the thinned image includes thinning a binarized image in which pixel values of the minute region pixels are inverted.
<7> Apply a smoothing filter to the skin luminance image,
Calculating a difference image between the skin luminance image after application of the smoothing filter and the skin luminance image;
Further including
The wrinkle analysis method according to any one of <1> to <6>, wherein the correlation image is generated from the difference image.
<8> The wrinkle index corresponding to the extracted average line segment pixel number based on the correlation between the average line segment pixel number extracted as the line segment length information and the wrinkle index value in which the wrinkle state is quantified. The wrinkle analysis method according to any one of <1> to <7>, further including calculating a value.
<9> By applying Fourier transform to the skin luminance image, the distribution information of the power spectrum for each angle is calculated,
From the distribution information, to calculate a directional entropy indicating the degree of bias of the distribution of the power spectrum for each angle,
The wrinkle analysis method according to any one of <1> to <8>, further including calculating a wrinkle index value from the line segment length information and the directional entropy.
<10> an image acquisition unit that acquires a skin luminance image of the analysis target part;
A pattern acquisition unit that acquires a binary pattern image representing a local wrinkle texture;
A correlation processing unit that generates a correlation image by sequentially calculating a correlation value between the binary pattern image and the skin luminance image while sliding the binary pattern image on the skin luminance image;
A binarization unit that generates a binarized image by binarizing the correlation image;
A thinning unit that generates a thinned image by thinning the binarized image;
An information extraction unit for extracting line segment length information from the thinned image;
An output processing unit that performs an output process of the line segment length information extracted by the information extraction unit;
Wrinkle analyzer.
<11> An angle correction unit that corrects an angle of the skin luminance image based on the directionality of wrinkles reflected in the skin luminance image;
Further including
The wrinkle analyzer according to <10>, wherein the correlation processing unit generates the correlation image from a skin luminance image that has been angle-corrected by the angle correction unit.
<12> An angle information acquisition unit that calculates power spectrum distribution information for each angle by applying Fourier transform to the skin luminance image,
Further comprising
The angle correction unit corrects the angle of the skin luminance image so that the directionality of wrinkles obtained from the distribution information matches the directionality of the local wrinkle texture represented by the binary pattern image, <11> Wrinkle analyzer described in 1.
<13> The pattern acquisition unit corresponds to the directionality of wrinkles reflected in the skin luminance image from a plurality of binary pattern images each representing a local wrinkle texture extending in the predetermined angle direction for each of a plurality of predetermined angles. The wrinkle analyzer according to <10>, which selects a binary pattern image.
<14> The pattern acquisition unit calculates power spectrum distribution information for each angle by applying Fourier transform to the skin luminance image, and corresponds to the distribution information from the plurality of binary pattern images. The wrinkle analyzer according to <13>, wherein a binary pattern image to be selected is selected.
<15> In the binarized image, a micro area removing unit that inverts a pixel value of a micro area pixel in a pixel having a pixel value representing a wrinkle texture,
Further comprising
The wrinkle analyzer according to any one of <10> to <14>, wherein the thinning unit thins a binarized image in which pixel values of the minute region pixels are inverted.
<16> a smoothing processing unit that applies a smoothing filter to the skin luminance image;
A difference calculating unit for calculating a difference image between the skin luminance image after application of the smoothing filter and the skin luminance image;
Further comprising
The wrinkle analyzer according to any one of <10> to <15>, wherein the correlation processing unit generates the correlation image from the difference image.
<17> A wrinkle index corresponding to the extracted average line segment pixel number based on a correlation between the average line segment pixel number extracted as the line segment length information and a wrinkle index value in which the wrinkle state is quantified. An index value calculation unit for calculating a value,
The wrinkle analyzer according to any one of <10> to <16>, further comprising:
<18> A distribution calculating unit that calculates power spectrum distribution information for each angle by applying Fourier transform to the skin luminance image;
From the distribution information, an entropy calculating unit that calculates directional entropy indicating the magnitude of the bias of the distribution of the power spectrum for each angle;
An index value calculation unit that calculates a wrinkle index value from the line segment length information and the directional entropy;
The wrinkle analyzer according to any one of <10> to <17>, further comprising:
<19> A program that causes at least one computer to execute the wrinkle analysis method according to any one of <1> to <9>.

2 CPU
3 Memory 4 Input / output I / F
DESCRIPTION OF SYMBOLS 7 Input part 9 Output part 10 Wrinkle analyzer 11 Image acquisition part 12 Pattern acquisition part 13 Correlation processing part 14 Binarization part 15 Thinning part 16 Information extraction part 17 Output processing part 21 Angle information acquisition part 22 Angle correction part 24 Smoothing Processing section 25 difference calculation section 26 minute area removal section 28 index value calculation section 31 distribution calculation section 32 entropy calculation section

Claims (11)

Acquire skin luminance image of the analysis target part,
Obtain a binary pattern image representing the local wrinkle texture,
While calculating the correlation value between the binary pattern image and the skin luminance image while sliding the binary pattern image on the skin luminance image, a correlation image is generated,
A binarized image is generated by binarizing the correlation image,
A thinned image is generated by thinning the binarized image,
Extracting line segment length information from the thinned image;
Wrinkle analysis method. Based on the directionality of wrinkles in the skin luminance image, the skin luminance image is angle-corrected.
Further including
The correlation image is generated by generating the correlation image from the angle-corrected skin luminance image.
The wrinkle analysis method according to claim 1. By applying Fourier transform to the skin luminance image, the distribution information of the power spectrum for each angle is calculated.
Further including
The angle correction is performed to correct the angle of the skin luminance image so that the directionality of wrinkles obtained from the distribution information matches the directionality of the local wrinkle texture represented by the binary pattern image.
The wrinkle analysis method according to claim 2. The acquisition of the binary pattern image corresponds to the directionality of wrinkles appearing in the skin luminance image from a plurality of binary pattern images respectively representing local wrinkle textures extending in the predetermined angle direction for each of a plurality of predetermined angles. Select a binary pattern image,
The wrinkle analysis method according to claim 1. By applying Fourier transform to the skin luminance image, the distribution information of the power spectrum for each angle is calculated.
Further including
The acquisition of the binary pattern image selects a binary pattern image corresponding to the distribution information from the plurality of binary pattern images.
The wrinkle analysis method according to claim 4. In the binarized image, the pixel value of the minute region pixel in the pixel having the pixel value representing the wrinkle texture is inverted.
Further including
The thinned image is generated by thinning a binarized image in which pixel values of the minute region pixels are inverted.
The wrinkle analysis method according to any one of claims 1 to 5. Applying a smoothing filter to the skin luminance image;
Calculating a difference image between the skin luminance image after application of the smoothing filter and the skin luminance image;
Further including
The correlation image is generated by generating the correlation image from the difference image.
The wrinkle analysis method according to any one of claims 1 to 6. The wrinkle index value corresponding to the extracted average line segment pixel number is calculated based on the correlation between the average line segment pixel number extracted as the line segment length information and the wrinkle index value in which the wrinkle state is quantified. To
The wrinkle analysis method according to any one of claims 1 to 7, further comprising: By applying Fourier transform to the skin luminance image, the power spectrum distribution information for each angle is calculated,
From the distribution information, to calculate a directional entropy indicating the degree of bias of the distribution of the power spectrum for each angle,
A wrinkle index value is calculated from the line segment length information and the directional entropy.
The wrinkle analysis method according to any one of claims 1 to 8, further comprising: An image acquisition unit for acquiring a skin luminance image of the analysis target part;
A pattern acquisition unit that acquires a binary pattern image representing a local wrinkle texture;
A correlation processing unit that generates a correlation image by sequentially calculating a correlation value between the binary pattern image and the skin luminance image while sliding the binary pattern image on the skin luminance image;
A binarization unit that generates a binarized image by binarizing the correlation image;
A thinning unit that generates a thinned image by thinning the binarized image;
An information extraction unit for extracting line segment length information from the thinned image;
An output processing unit that performs an output process of the line segment length information extracted by the information extraction unit;
Wrinkle analyzer.   A program for causing at least one computer to execute the wrinkle analysis method according to any one of claims 1 to 9.