JP2005137756A - Health measuring system and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a health measuring technique easily and objectively determining comprehensive health based on information of a part of the body. <P>SOLUTION: This health measuring system extracts a region of the tongue from an image F1 of the tongue TN obtained with a digital camera. This system measures the colors of the base SJ and nipples NT as the color of the tongue and the fur (feature quantities) to be diagnosed by the glossoscopy based on the image of the tongue region and similarly measures the shape of the tongue, a teeth mark, a tear mark, wetness of the fur, smoothness of the fur, the thickness of the fur and a pattern on the fur as the feature quantities. Information of the nine types of feature quantities is reduced into one Mahalanobis distance information used as an index of the health. The Mahalanobis distance determines the health by comparing a distance between a reference space formed using a healthy subject as a sample and a consulter. Consequently, this system can easily and objectively determine the comprehensive health based on the Mahalanobis distance (output information) reflecting the multiple feature quantities obtained from the tongue image (the image of the part of the body). <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a health level measurement technique for measuring a health level based on an image of a captured body part.

  In Oriental medicine, there is a diagnostic method (tongue examination) for diagnosing a health condition or a medical condition by observing the state of the tongue, which is a body part.

  This tongue diagnosis is disclosed in, for example, Patent Document 1, and it is useful to compare the tongue image of the patient with the past standard tongue image (standard tongue image) and compare the color of the tongue, etc. Describes diagnosis support technology that can make a diagnosis without relying on it.

Japanese Patent No. 2763989

  In the technique of the above-mentioned Patent Document 1, the determination is based on the experience and know-how of a doctor, and the determination criterion is not quantitative. Therefore, individual differences occur and objective determination is difficult. In addition, there are many items for diagnosis in tongue examination, and since the determination criteria and weighting of each are complicated and difficult to understand, it is not easy to determine the overall health level.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a health level measurement technique that can easily and objectively determine the overall health level based on information on a body part.

  In order to solve the above problems, the invention of claim 1 is a health degree measurement system, wherein (a) an imaging means for imaging a body part and acquiring a photographed image relating to the body part; ) Measuring means for measuring n types (n is an integer of 2 or more) of feature quantities related to the body part based on the captured image; and (c) performing predetermined processing using the n types of feature quantities as input information. Processing means for generating m types (m is a natural number) of output information smaller than the n types, and the m types of output information are used to determine the health level of the person having the body part.

  According to a second aspect of the present invention, in the health level measuring system according to the first aspect of the invention, the body part is a tongue.

  Further, the invention of claim 3 is the health degree measuring system according to claim 2, in which the n kinds of feature values are tongue color, tongue shape, tooth trace, crest, moss color, moss moisture. And at least one feature selected from the group consisting of moss smoothness, moss thickness, and moss pattern.

  According to a fourth aspect of the present invention, in the health level measuring system according to the first aspect of the invention, the body part is a face.

  According to a fifth aspect of the invention, in the health measurement system according to the first aspect of the invention, the body part is an eye.

  Further, the invention of claim 6 is the health degree measurement system according to any one of claims 1 to 5, wherein the n types of feature amounts are a luminance distribution and a chromaticity acquired from the photographed image. And at least one feature quantity selected from the group consisting of a distribution, a histogram shape, a contour, and a texture.

  According to a seventh aspect of the present invention, in the health measurement system according to any one of the first to sixth aspects, the m types of output information are one type of output information.

  According to an eighth aspect of the present invention, in the health level measuring system according to the seventh aspect of the invention, the one type of output information is information on the Mahalanobis distance based on a feature amount related to a part of a healthy body.

  The invention of claim 9 is the health degree measurement system according to any one of claims 1 to 8, wherein the imaging means (a-1) emits light of three or more wavelength bands different from each other. It has a filter to transmit.

  Further, the invention of claim 10 is the health level measurement system according to any one of claims 1 to 9, wherein (d) a light emission amount with a constant illuminance of a body part imaged by the imaging means Further provided is an illumination means capable of controlling the above.

  The invention according to claim 11 is executed by a computer included in the health measurement system, thereby causing the health measurement system to function as the health measurement system according to any one of claims 1 to 10. It is a program.

  According to the first to eleventh aspects of the present invention, predetermined calculation is performed using n types of feature quantities measured based on a captured image obtained by imaging a body part as input information, and m types of output information less than n types are output. Therefore, it is possible to easily and objectively determine the overall health level based on the body part information.

  In particular, in the invention of claim 3, the n types of feature quantities are tongue color, tongue shape, tooth marks, fissures, moss color, moss dampness, moss smoothness, moss thickness, and moss. Since at least one feature amount selected from the group consisting of patterns is included, output information can be generated appropriately.

  In the invention of claim 6, the n types of feature quantities are at least one selected from the group consisting of luminance distribution, chromaticity distribution, histogram shape, contour, and texture acquired from the captured image. Since the feature amount is included, output information can be generated appropriately.

  In the invention of claim 7, since the m types of output information are one type of output information, the overall health level can be determined more easily.

  In the invention according to claim 8, since one type of output information is information on the Mahalanobis distance based on the feature amount relating to the part of the healthy body, the health degree can be accurately determined.

  Further, in the invention of claim 9, since the filter for transmitting light of three or more wavelength bands different from each other is provided, it is possible to improve the measurement accuracy of the feature amount related to the body part.

  Further, in the invention of claim 10, since the illumination means capable of controlling the light emission amount with the illuminance of the body part imaged by the imaging means constant is provided, the measurement accuracy of the feature quantity relating to the body part can be improved. .

<Main components of the health measurement system>
Before describing the configuration of the health level measurement system, first, the health level measurement method will be described below.

  Methods for measuring health include diagnoses that can be easily performed at home, such as temperature measurement, blood pressure, and pulse, as well as blood tests, urinalysis, electrocardiograms, X-rays, and other diagnoses that are performed in the workplace and local governments, as well as medical treatment such as CT and MRI. There are advanced inspections carried out by institutions. However, these are for evaluating one aspect of the state of the body or examining the presence or absence of a specific disease, and are not a comprehensive diagnosis of health.

  On the other hand, oriental medicine has a diagnostic method called tongue diagnosis, and it is possible to judge the constitution and bias of physical condition from the color and shape of tongue and moss. The tongue is said to be a mirror that easily shows changes in the general condition of the whole body. In other words, the tongue tissue is rich in blood vessels, and the surface has no stratum corneum and is covered with mucous membrane, so its color reflects the color of blood and body fluids and is not affected by sunburn or race There is an advantage.

  A specific configuration of a health measurement system that measures health using the above tongue examination will be described below.

  FIG. 1 is a schematic diagram illustrating a main configuration of a health degree measurement system 1 according to an embodiment of the present invention.

  The health level measurement system 1 includes a camera 2 that captures the tongue TN of the examinee, and a processing device 3 that is communicably connected to the camera 2 via a cable CB.

  The camera 2 is configured as, for example, a digital camera. A display unit 20 such as a liquid crystal display is provided on the back surface, and a two-lamp type flash 21 (21a, 21b) fixed to the tip of an arm extending from the lens barrel. Is provided.

  The flash 21 irradiates the tongue TN with illumination light at a specific color temperature (spectral characteristic). Further, the flash 21 can control the light emission amount based on the distance information to the tongue TN so that the illuminance of the tongue TN to be imaged is constant. Thereby, the luminance information of the tongue TN that is the subject of photographing can be accurately acquired.

  The processing device 3 is configured as a personal computer, for example, and can process image data. The processing device 3 includes a processing unit 30 having a box-like shape, an operation unit 31, and a display unit 32 composed of, for example, a CRT.

  A drive 301 for inserting the optical disk 9 is provided on the front surface of the processing unit 30.

  The operation unit 31 includes a mouse 311 and a keyboard 312 and receives an operation input to the processing device 3.

  FIG. 2 is a diagram illustrating functional blocks of the health degree measurement system 1.

  The camera 2 includes an imaging unit 22 that captures an image of the tongue TN via the optical unit 21 and a control unit 23 that is connected to these units so as to be able to transmit. The camera 2 includes a communication unit I / F 24 that is connected to the control unit 23 so as to be able to transmit.

  The optical unit 21 includes a lens group and a drive unit that changes the arrangement, and focuses and zooms to form an optical image of the tongue TN on the imaging unit 22.

  The imaging unit 22 includes, for example, an imaging element such as a CCD in which three-band (RGB) filters are arranged in a Bayer array, and generates RGB image signals. That is, the imaging unit 22 has filters that transmit light in three different wavelength bands (RGB), and functions as an imaging unit that captures the tongue TN and acquires a tongue image. After the analog image signal of the tongue TN of the examinee is acquired by the imaging unit 22, A / D conversion is performed to generate digital signal image data.

  The communication I / F 24 is an interface for data transmission with the processing device 3 via the cable CB.

  The control unit 23 has a CPU and is a part that performs overall control of the operation of the camera 2. In addition, the control unit 23 performs control for transmitting the image data acquired by the imaging unit 22 to the processing device 3 via the communication I / F 24.

  The processing device 3 includes a control unit 33 that is connected to the operation unit 31 and the display unit 32 so as to be able to transmit. In addition, the processing device 3 includes a storage unit 34 that is connected to the control unit 33 so as to be able to transmit, an image processing unit 35, an input / output I / F 36, and a communication I / F 37.

  The storage unit 34 is configured as a hard disk, for example, and stores a health degree measurement program DP described later.

  The image processing unit 35 is a part that performs image processing on image data captured by the camera 2.

  The input / output I / F 36 is an interface for inputting / outputting data to / from the optical disk 9 as a recording medium via the drive 301.

  The communication I / F 37 is an interface for data transmission with the camera 2 via the cable CB.

  The control unit 33 includes a CPU 331 and a memory 332 that function as a computer, and is a part that performs overall control of the operation of the processing device 3. By executing the health degree measurement program DP in the control unit 33, an operation for measuring the health degree of the examinee whose tongue TN has been photographed is performed in the processing device 3 (detailed later).

  Program data such as the health support program DP recorded on the optical disc 9 can be stored in the memory 332 of the control unit 33 via the input / output I / F 36. As a result, the stored program can be reflected in the operation of the processing device 3.

<About the operation of the health measurement system 1>
FIG. 3 is a flowchart showing the basic operation of the health degree measurement system 1. This operation is performed by executing the health degree measurement program DP in the processing device 2 by the control unit 33.

  First, the tongue region is extracted by the processing device 3 from the image data of the tongue TN acquired by the camera 2 (step S1).

  In step S2, the color (characteristic amount) related to the tongue TN is measured.

  In step S3, color coordinates are converted for the color measured in step S2.

  In step S4, other feature amounts are measured.

  In step S5, the Mahalanobis distance is calculated based on the feature values measured in steps S2 and S4.

  The operation as described above is performed in the health degree measurement system 1, but the operation of steps S1 to S5 will be described below.

<1. Extraction of tongue area>
In the image processing unit 35 of the processing device 3, the tongue region is extracted from the image data of the tongue TN photographed by the camera 2. This extraction process will be described below.

  First, the photographed image F1 (FIG. 4) is acquired by the camera 2 so that the lower half of the face is photographed around the tongue TN. In this case, the photographed image F1 includes the face surface, lips, teeth and the like in addition to the tongue TN.

  Therefore, the region of the tongue TN is extracted based on the luminance edge of the captured image F1. Specifically, the difference between four neighborhoods (up, down, left, and right) is taken in the G data of each pixel, and pixels whose difference value (luminance difference) exceeds a predetermined threshold value α are extracted to determine the edge of the tongue region. In this case, there is a luminance difference around the tongue TN due to the mouth, teeth, shadow of the tongue, etc., and a threshold value α of the luminance difference that enables detection of the tongue TN is set separately from these. Thereby, only the area | region of tongue TN can be extracted from the picked-up image F1.

<2. Color measurement>
Based on the tongue region extracted from the photographed image F1, the color relating to the tongue TN is measured. This color measurement will be described below.

  An image F2 in FIG. 4 is an enlarged image of the surface of the tongue TN. The tongue TN is covered with mucous membranes, and many filamentous nipples (hereinafter, also simply referred to as “nipples”) NT are distributed almost on the entire surface of the tongue. Since this nipple NT has no blood vessels, it usually looks translucent to white. On the other hand, since the base SJ (parallel hatched portion) other than the nipple NT is rich in capillaries, the color of blood appears and appears red.

  Such nipple NT and base SJ are subjected to area discrimination by the image processing unit 35 using the characteristics of the colors. That is, since the teat NT is white, the RGB values are almost equal, but the base SJ is red, so the R value is particularly large compared to G. Therefore, the ratio of R and G of each pixel, for example, R / G, is calculated in the captured tongue image data. Then, the red region exceeding the predetermined threshold value β is determined as the base SJ portion, and the other white region is determined as the nipple NT portion. This threshold value β is set in consideration of the color temperature of the flash 21 and the filter characteristics of the camera 2.

  The processing device 3 measures the colors related to the nipple NT and the base SJ whose areas have been discriminated as described above. Here, the moss on the tongue TN is formed by keratinization of the nipple NT of the tongue mucosa and adhesion of detached cells, mucus, food, and bacteria, so the measured color of the nipple NT becomes the color of moss. On the other hand, the portion of the tongue TN other than the nipple NT, that is, the measured color of the base SJ becomes the ground color of the tongue.

  As described above, the color of the tongue and moss, which are measurement items (features) used in the tongue diagnosis, are measured.

<3. Conversion of color coordinates>
Color coordinate conversion is performed on the measured tongue and moss colors. This color coordinate conversion will be described below.

  The captured image acquired by the camera 2 is formed as unique RGB data reflecting the illumination (flash 21) and the spectral characteristics of the camera 2. Therefore, the RGB data is converted into XYZ (Yxy) chromaticity coordinates standardized by, for example, the International Lighting Commission using a regression equation.

Specifically, first, a plurality of color charts (color charts) with known XYZ chromaticity values are photographed in a state where the flash 21 is caused to emit light, and RGB data of the photographed image is recorded. Then, from the relationship between the RGB data of the captured image and the color chart, the elements a11 to a33 of the 3 × 3 matrix (conversion matrix) of the following equation (1) are obtained.

Here, since a plurality of combinations of the XYZ values of the color chart and the RGB values of the camera are obtained, the XYZ chromaticity values of the color chart are obtained from the RGB data of the photographed image by regression analysis using the former as the objective variable and the latter as the explanatory variable. A transformation matrix is obtained.

  Next, the XYZ chromaticity values are calculated by substituting the RGB data of the tongue and moss obtained by actually photographing the tongue TN into the right side of the equation (1) and multiplying by the already obtained conversion matrix. . By such color coordinate conversion, an absolute index that excludes spectral characteristics such as illumination light with respect to tongue and moss colors can be obtained. And in calculation of Mahalanobis distance in step S5, this XYZ value is utilized as input information.

<4. Measurement of other features>
In the above, the color of tongue and moss, which are two types of feature values related to the tongue, was measured. As other feature values (measurement items), (1) tongue shape, (2) tooth trace, (3) fissure, (4) Moisture of moss, (5) Smoothness of moss, (6) Thickness of moss, (7) Measure the pattern of moss. These seven types of feature amounts will be described in order below.

<4-1. Tongue shape>
The shape of the tongue TN changes due to excess or deficiency of blood or body fluid, and becomes enlarged or thinned. When the tongue is enlarged, the outline of the tongue swells round like the tongue TNa of FIG. 5 (a), whereas when the tongue thins, the outline of the tongue like the tongue TNb of FIG. 5 (b). Becomes close to a triangle. Therefore, the shape of the tongue is measured based on the contour line of the tongue TN extracted in step S1 of FIG.

  Specifically, first, the coordinates of the center of gravity GD of the tongue TN are obtained based on the average value of the coordinates of all the pixels constituting the tongue region (FIG. 6). Next, distances L1, L2,..., Ln from the center of gravity GD to the contour are obtained at equal angles clockwise starting from a straight line extending in the vertical direction from the center of gravity GD of the tongue TN. Then, using the obtained distances L1 to Ln, the variance of the distance is calculated based on the following equation (2).

  By calculating the dispersion in this way, it can be determined that the tongue TN is enlarged and close to a circle when the dispersion value is small, and that the tongue TN is thin and close to a triangle when the dispersion value is large. I can judge. In the calculation of the Mahalanobis distance in step S5 in FIG. 3, this variance value is used as input information.

<4-2. Tooth marks>
The tooth mark on the tongue is generated when the amount of water is excessive, and the peripheral edge (contour) of the tongue TNc in FIG. 7 is uneven. Therefore, the presence / absence of the tooth trace and the degree of the tooth trace are measured by judging the unevenness of the contour line of the tongue extracted in step S1 of FIG.

  Specifically, as shown in FIG. 6, using the distances L1 to Ln from the center of gravity GD of the tongue to the contour, the difference between two adjacent distances is obtained based on the following equation (3), and the sum is calculated. To do.

  By calculating the sum of the differences in this way, if the difference sum is large, it can be determined that there is a tooth mark and the degree is large, and if the difference sum is small, there is almost no tooth mark and the degree. Can be judged to be small. Then, in the calculation of the Mahalanobis distance in step S5 in FIG. 3, the sum of the differences is used as input information.

<4-3. Crest>
Tongue tears occur when mucosal regeneration is reduced due to malnutrition or the like, and groove-like tears RM are formed mainly at the center of the tongue like the tongue TNd in FIG. 8 (a). Since the tear pattern RM does not have the nipple NT, the color of this portion is the color of the base SJ. Therefore, the tear pattern RM formed as the base region at the center is detected using the threshold value β used for discriminating between the nipple NT and the base SJ.

  Specifically, in order to distinguish from map-like (ie, block-like) moss, etc., as shown in FIG. 8B, the pixels having the color of the background SJ are searched up and down and left and right, and the connected pixels are counted. A portion where the number of connected pixels is equal to or greater than a predetermined value and the other number of connected pixels is equal to or smaller than a predetermined value is determined as a groove, that is, a tear pattern RM.

  By performing such a pixel search, when the number of connected pixels increases, it can be determined that a large tear pattern RM is formed. When the number of connected pixels is small, there is no crack pattern RM, or Can be judged small. In the calculation of the Mahalanobis distance in step S5 in FIG. 3, the number of connected pixels is used as input information.

<4-4. Moistness of moss>
The wetness of the moss varies depending on the moisture state of the tongue TN, and the moss portion on the surface of the central portion is mainly wetted or dried. When the surface of the moss is wet, a portion WT of an image is generated in which the illumination light from the flash 21 is regularly reflected and flies white like the tongue TNe in FIG. And since the surface of the tongue is curved and the unevenness is formed on the surface by the nipple, the pixels that fly white are limited. Therefore, the wetness of the tongue is measured by detecting white-out pixels having a certain brightness and counting the number of pixels.

  Specifically, a pixel having an xy chromaticity value of, for example, 0.3 ≦ x and y ≦ 0.35 is estimated as white, and a pixel having a Y value of, for example, 90 ≦ Y (maximum luminance value = 100) is estimated. Estimated bright.

  By detecting whiteout pixels in this manner, it can be determined that the moss is wet when there are many whiteout pixels, and it can be determined that the moss is dry when there are few whiteout pixels. Then, in the calculation of the Mahalanobis distance in step S5 in FIG. 3, the number of whiteout pixels is used as input information.

<4-5. Smoothness and thickness of moss>
The smoothness and thickness of the moss vary depending on the amount of deposits on the teat NT. That is, when there are few deposits, since the moss is thin and the nipple NT is seen independently, it feels rough. On the other hand, when there are many deposits, the moss is thick and the nipple NT covers the surface of the tongue. Therefore, the smoothness and thickness of the moss are measured based on the ratio of the number of pixels relating to the teat NT and the base SJ determined in step 2 of FIG.

  Specifically, a region near the center of the tongue to which moss adheres is extracted, and the pixel of the nipple NT and the pixel of the base SJ are discriminated by the color threshold β. Then, the ratio of the number of pixels between the nipple NT and the base SJ is calculated.

  By calculating the ratio of the number of pixels in this way, when the ratio of the background SJ is high, it can be determined that the moss is thin and rough, and when the ratio of the nipple NT is high, the moss is thick and sticky. Can be judged. In the calculation of the Mahalanobis distance in step S5 in FIG. 3, the ratio of the number of pixels between the nipple NT and the base SJ is used as input information.

<4-6. Moss pattern>
When a part of the moss is peeled off due to malnutrition or the like, a mottled pattern is formed as in the tongue TNf in FIG. In this pattern, the peeled area HK is the color of the base SJ, and the area other than the peeled area HK is the color of the nipple NT.

  Therefore, as shown in FIG. 10B, the central portion of the tongue TNf is divided into a plurality of slightly larger rectangular blocks BK, and the peeling area HK is determined using the average value of the colors in each rectangular block BK. Here, the rectangular block BK corresponding to the peeling area HK is detected using the threshold value β that identifies the nipple NT and the base SJ.

  By detecting such a rectangular block BK, when there are many rectangular blocks BK corresponding to peeling, it can be determined that the mottled pattern is remarkable, and when there are no or few rectangular blocks BK corresponding to peeling. Can be determined to have no mottled pattern or a low degree of pattern. In the calculation of the Mahalanobis distance in step S5 in FIG. 3, the rectangular block BK corresponding to the peeling is used as input information.

<5. Calculation of Mahalanobis distance>
In the above, we measured the tongue and moss color, tongue shape, tooth marks, clefts, moss dampness, moss smoothness, moss thickness, moss pattern as tongue features (measurement items), These nine types of feature amounts are used as input information to perform predetermined processing, and one type of output information called Mahalanobis distance based on the feature amount relating to the tongue of the healthy body is generated. Here, the Mahalanobis distance is a distance scale calculated in consideration of variations and correlations for many measurement items, and is expressed as the following equation (4).

  In the above equation (4), k represents the number of measurement items such as tongue and moss color, and m represents the average value of measurement data for each measurement item. Further, σ represents a standard deviation, and aij represents an element of an inverse matrix with respect to a correlation matrix representing a correlation between measurement items.

  Here, when obtaining the distance function shown in Expression (4), measurement data of each item measured from an image obtained by photographing a healthy person's tongue TN is used as a reference data group. Then, the average of the Mahalanobis distance based on measurement data obtained by sampling a healthy person is set to 1.

  Thus, the health degree of the examinee is determined based on the Mahalanobis distance D calculated by substituting each measurement data (feature amount) obtained from the image obtained by photographing the examinee's tongue TN into the above equation (4). it can. That is, if the Mahalanobis distance D is small, it can be determined that the examinee is close to the reference space and is healthy. If the Mahalanobis distance D is large, the examinee is far from the reference space and the examinee has some health problem such as a disease. To be judged. As a specific example, when the Mahalanobis distance D is 2 to 5, the physical condition is slightly worse, and when the Mahalanobis distance D exceeds 5, it is diagnosed that a hospital visit is necessary.

  In order to calculate the Mahalanobis distance D of 1 from the many features obtained from the tongue image by the operation of the health level measuring system 1 as described above, the overall health level determination for the person with the photographed tongue is performed. It will be easy and objective.

<Modification>
◎ In the above embodiment, it is not essential to use measurement data (features) of diagnostic items used in tongue examination for health level measurement. The frequency distribution of the luminance level of each color in the tongue image and the frequency of the tongue image Features such as components, distribution status for each predetermined region in the tongue image, and changes in the state of the tongue over time may be used.

  A technique for identifying an image using a feature amount extracted from the image is widely known in the field of searching for an image on the Internet or the like (for example, JP-A-5-242160). In this technique, the similarity between images is expressed numerically using features such as color and shape, and this is applied to the tongue image. A specific example will be described below.

  As shown in FIG. 11A, an image is cut out from a square region RC inscribed in the tongue TN, and the feature amount of this image is extracted. Here, the extracted feature quantities are luminance and color distribution, histogram shape, contour and texture.

  Specifically, as shown in FIG. 11 (b), the cut out square area RC is divided into four parts in the vertical and horizontal directions to generate 16 blocks BR. Next, as shown in FIG. 12, a histogram of luminance and color is created. Here, RGB pixel values for each block BR are color-coordinate-converted into Yxy values, and histograms relating to luminance Y and chromaticity x, y are created as shown in FIGS. Then, the average value, maximum value, minimum value, and peak value on the horizontal axis of each histogram are detected, and the histogram is divided into four equal parts in the horizontal axis direction as in regions P1 to P4 shown in FIG. The sum of the level values is calculated for each of the areas P1 to P4. Thereby, the distribution of luminance and color and the shape of the histogram can be acquired as the feature amount of the image.

  Further, a differential filter shown in FIG. 13 is applied to the cut-out square region RC to calculate the sum. Thereby, the outline as the feature amount of the image can be acquired.

  Further, the extracted square region RC is Fourier transformed, and the transformation result is divided into four equal parts with respect to the frequency (FIG. 14). And the sum total of the intensity | strength regarding each level Q1-Q4 divided into 4 equal parts is calculated. Thereby, the texture as the feature amount of the image can be acquired.

  Note that the above-mentioned numerical values of the diagnostic items of the tongue diagnosis are indirectly included in the above-described image feature amount, such as a difference in brightness between the central region and the peripheral region due to shadows around the enlarged tongue. ing. On the other hand, in the medical field, features that are not recognized as diagnostic items can also be used as diagnostic items.

  As described above, it is possible to appropriately measure the health level by extracting the feature values from the tongue image and utilizing these feature values for the health level measurement.

  The health level measurement described above is not limited to the feature amount extracted from the tongue image, but may use a feature amount extracted from a captured image obtained by photographing a body part such as a face or an eye. .

  In other words, the fact that the face color represents a health condition is experienced in daily life, and is used for diagnosis as one of the desired medical examinations in Oriental medicine. Specifically, when the complexion is white, it is conceivable that the coldness and blood volume decrease, and when it is blue-purple, the coldness and blood circulation are decreased. If the complexion is red, fever and water shortage may occur. If it is yellow, digestive system malfunction may occur. If it is black, liver or kidney malfunction may occur. In addition, since a capillary network is developed in the skin of the face, blood vessel contraction and expansion, some blood circulation, blood components, etc. are likely to appear as color changes. In particular, the skin is thin on both cheeks and around the eyes, and color changes are likely to appear.

  Therefore, the camera captures the face and divides it into areas such as the forehead, cheeks, around eyes / nose / mouth, and chin in the image of the face. Then, feature quantities such as luminance and color distribution, histogram shape, contour and texture are measured in each of the divided areas, and the Mahalanobis distance is obtained from these feature quantities. As a result, the health degree can be measured from the face image as in the case of the tongue image.

  In addition, iris diagnostics in which organ abnormalities appear in the iris through the optic nerve have been studied for a long time in Europe and the United States. Various charts of diseases of various organs such as stomach, liver, gallbladder, duodenum, and heart and corresponding iris positions are provided as charts (see, for example, JP-A-2001-128958).

  Therefore, the eye is photographed with a camera, and the pupil and iris are divided into a plurality of regions in the radial direction and the circumferential direction in the photographed image of the eye. Then, feature quantities such as luminance and color distribution, histogram shape, contour and texture are measured in each of the divided areas, and the Mahalanobis distance is obtained from these feature quantities. As a result, the degree of health can be measured from the eye image as in the case of the tongue image.

  In the above embodiment, it is not essential to measure the health level with a fixed measurement item, and the measurement item may be changed according to the type of disease or the like to be examined.

  According to Oriental Medicine, digestive diseases are likely to appear in the color and shape of moss, and circulatory diseases are likely to appear in the color and shape of the tongue. Therefore, the maharanobis distance is calculated by selecting measurement items centered on moss for diagnosis of digestive diseases and selecting measurement items centered on tongue for diagnosis of circulatory diseases. By doing so, the measurement accuracy of the health level can be improved.

  Similarly, you may change the sample for producing | generating the reference | standard space regarding Mahalanobis distance for every disease which wants to consult. In other words, the gastrointestinal healthy person is sampled for gastrointestinal diseases, and the heart space is sampled for circulatory disease to generate a reference space, thereby improving the measurement accuracy of the health level. It can be improved.

  When changing the above measurement items and samples, care should be taken to maximize the difference in Mahalanobis distance between healthy and sick people.

  In the above embodiment, the measurement data of healthy persons for generating the reference space related to the Mahalanobis distance and the measurement items may be updated as new knowledge increases.

  Since the Oriental medicine tongue examination is based on experience, there is a possibility that a new item for measuring the health level may be discovered. When a new disease occurs, the types of diseases that can be diagnosed can be increased by setting measurement items and samples that maximize the Mahalanobis distance of the patient. Then, every time a measurement item or reference space is changed or added, the health degree determination accuracy can be improved by updating the health degree measurement program DP in the storage unit 34 of the processing device 3 or the database regarding the sample. In addition, the business operator who sells the health level measurement system 1 can also make a profit by providing not only the sales but also the health level measurement program DP and the database update service.

  In the imaging unit 22 in the above embodiment, it is not essential to have RGB 3-band filters, and it may have filters of 4 bands or more.

  FIG. 15 is a diagram showing an image sensor 4 according to a modification of the present invention.

  The imaging element 4 is provided with a lens array 42 and a filter array 43 on the front surface of the imaging sensor 41. The filter array 43 is divided into nine regions, and each region has characteristics of different transmission wavelengths. That is, the filter array 43 functions as a filter that transmits light in nine different wavelength bands. A spectral image of 9 bands can be acquired by the imaging device 4 having such a configuration.

  Then, when estimating the XYZ values by regression, if the number of bands is photographed as in the imaging device 4 described above, the explanatory variables will increase, and the accuracy will be improved.

  In the above embodiment, it is not essential to reduce n kinds of measurement information such as nine kinds to one kind of index called Mahalanobis distance, and it is reduced to m kinds (m <n) of indices representing health. May be. For example, by generating two pieces of output information of the digestive system and the circulatory system, a multifaceted and comprehensive health level can be grasped.

  In the above embodiment, it is not essential to calculate the XYZ value by multiplying the three RGB inputs and the 3 × 3 transformation matrix as shown in Equation (1). As shown, an XYZ value may be calculated by multiplying 6 inputs to which secondary terms (R × B, R × G, G × B) are added and a 3 × 6 transformation matrix.

  Thereby, the accuracy of the calculated XYZ value is improved.

  Further, by calculating the measurement data by photographing the actual tongue instead of the color chart (color chart) described above, the calculation accuracy of the XYZ values can be improved.

  In the above embodiment, it is not essential to configure the health measurement system separately for the camera and the processing device, and the function of the processing device may be provided on the camera side. Thereby, the health degree measured by photographing the tongue with the camera alone can be displayed on the display unit 20 (FIG. 1) of the camera, and the convenience for the examinee is improved.

It is the schematic which shows the principal part structure of the health degree measurement system 1 which concerns on embodiment of this invention. It is a figure which shows the functional block of the health degree measurement system. 3 is a flowchart showing a basic operation of the health level measurement system 1. It is a figure for demonstrating a tongue image. It is a figure for demonstrating the shape of a tongue. It is a figure for demonstrating the measurement of the shape of a tongue. It is a figure for demonstrating a tooth trace. It is a figure for demonstrating a crack. It is a figure for demonstrating the measurement of the moisture of moss. It is a figure for demonstrating the pattern of moss. It is a figure for demonstrating extraction of the feature-value which concerns on the modification of this invention. It is a figure for demonstrating extraction of a feature-value. It is a figure for demonstrating extraction of a feature-value. It is a figure for demonstrating extraction of a feature-value. It is a figure which shows the image pick-up element 4 which concerns on the other modification of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Health degree measurement system 2 Camera 3 Processing apparatus 21, 21a, 21b Flash 22 Imaging part 33 Control part 35 Image processing part DP Health degree measurement program NT Papilla SJ Base TN, TNa-TNf Tongue

Claims (11)

A health measurement system,
(a) imaging means for imaging a body part and acquiring a captured image relating to the body part;
(b) a measurement unit that measures n types (n is an integer of 2 or more) of feature amounts related to the body part based on the captured image;
(c) processing means for performing predetermined processing using the n types of feature values as input information and generating m types (m is a natural number) of output information less than the n types;
With
The m-type output information is used for determining a health level of a person having the body part. The health degree measurement system according to claim 1,
The health measuring system, wherein the body part is a tongue. In the health level measurement system according to claim 2,
The n types of feature quantities are selected from the group consisting of tongue color, tongue shape, tooth marks, fissures, moss color, moss moisture, moss smoothness, moss thickness, and moss pattern. A health level measurement system including at least one feature amount. The health degree measurement system according to claim 1,
The health measuring system, wherein the body part is a face. The health degree measurement system according to claim 1,
The health measurement system, wherein the body part is an eye. In the health measurement system according to any one of claims 1 to 5,
The n types of feature amounts include at least one feature amount selected from the group consisting of luminance distribution, chromaticity distribution, histogram shape, contour, and texture acquired from the captured image. Health level measurement system. The health degree measurement system according to any one of claims 1 to 6,
The m-type output information is one type of output information. The health measurement system according to claim 7,
The health information measuring system according to claim 1, wherein the one type of output information is Mahalanobis distance information based on a feature amount related to a part of a healthy body. The health degree measurement system according to any one of claims 1 to 8,
The imaging means includes
(a-1) a filter that transmits light of three or more different wavelength bands,
A health measurement system characterized by comprising: The health measurement system according to any one of claims 1 to 9,
(d) Illumination means capable of controlling the amount of emitted light to make the illuminance of the body part imaged by the imaging means constant,
A health measurement system further comprising:   A program for causing a health degree measurement system to function as the health degree measurement system according to any one of claims 1 to 10, when executed on a computer included in the health degree measurement system.

Images