Human body back acupoint recognition method and device, storage medium and program product

Abstract

The embodiment of the invention provides a human body back acupoint recognition method and device, a storage medium and a program product, and belongs to the technical field of acupoint recognition, the method comprises the steps that a to-be-recognized human body back image is acquired, and positioning icons are preset in the Dazhui acupoint and the lumbosacral vertebrae in the back image; respectively identifying positioning icons of a Dazhui point and a lumbosacral vertebrae in the back image; central points of two positioning icons in the back image are determined, and the central points of the two positioning icons correspond to the corresponding Dazhui point and lumbosacral vertebrae respectively; according to the center points of the two positioning icons, the space between the Dazhui point and the lumbosacral vertebrae is equally divided into a plurality of points. According to the embodiment of the invention, the multiple points between the two acupuncture points are determined according to the preset positioning icons of the Dazhui acupuncture point and the lumbosacral vertebrae in the human body back image, so that the related acupuncture points of the back are massaged, the upper part of the human body does not need to be exposed, the privacy of the user is protected, and the user experience is improved.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the technical field of acupoint identification, and in particular, to a method, device, computer-readable storage medium, and computer program product for identifying acupoints on the back of a human body. Background Art

[0002] In robot massage or intelligent massage, accurate identification of human acupuncture points is a prerequisite for the effectiveness of massage.

[0003] With the development of society, various methods for predicting the human back contour and acupoint recognition have emerged in an endless stream. The common methods for recognizing the human back contour and acupoints are as follows: 1. Human back contour recognition technology: Human back contour recognition mainly relies on computer vision and image processing technology. These technologies achieve accurate recognition of the back contour by preprocessing the back image, extracting features, and detecting contours. Specifically, they include: 1. Image preprocessing: including denoising, contrast enhancement, grayscale, and other steps to improve image quality and provide a good foundation for subsequent processing. 2. Feature extraction: Use edge detection, corner detection, and other methods to extract key feature points of the back, such as the contour information of the shoulder blade, spine, and other parts. 3. Contour detection: Accurately extract and describe the back contour through contour tracking algorithms or level set-based methods. 2. Human back acupoint recognition technology: Human back acupoint recognition technology is more complicated and requires a comprehensive analysis combined with traditional Chinese medicine meridian theory and modern scientific and technological means. At present, there are mainly the following methods: 1. Acupoint recognition based on image processing: ① Image registration: The collected back image is registered with the standard acupoint map to determine the exact location of the acupoint. ② Feature point matching: The location of the acupoint is identified by extracting the feature points in the image and matching them with the known acupoint features. 2. Acupoint recognition based on machine learning: ① Training model: Using a large amount of labeled data, including back images and corresponding acupoint locations, the machine learning model is trained to enable it to automatically identify acupoints. ② Prediction and verification: The back image to be identified is input into the trained model to predict the acupoint location, and the accuracy of the recognition is ensured through the verification step. 3. Acupoint recognition based on deep learning: ① Deep learning network: Using deep learning technology, such as convolutional neural network, a complex neural network model is constructed to extract and classify features of the back image. ② End-to-end recognition: The end-to-end recognition process from image input to acupoint location output is realized to improve the accuracy and efficiency of recognition. 4. Sensor-based acupoint positioning: ① Pressure sensor: By placing a pressure sensor array on the back, the pressure changes in different parts are detected to determine the location of the acupoints. ② Electromagnetic sensor: Using the electromagnetic principle, the acupoints are located by measuring the changes in the electromagnetic field in different parts of the back.

[0004] However, these technologies have some shortcomings and deficiencies, which are mainly reflected in the following aspects: I. Technical limitations: 1. Recognition accuracy: Although relevant technologies have made certain progress, in some cases, such as large individual differences and severe environmental interference, the recognition accuracy may still be limited. This may lead to inaccurate acupoint positioning and affect the treatment effect. 2. Scope of application: Current technology may not be applicable to all types of back contours and acupoints. For example, for individuals with certain special body shapes or pathological conditions, the recognition effect may not be good. II. Individual differences: 1. Physiological structure differences: There are differences in the back contours and acupoint locations of different individuals, which makes it difficult for recognition technology to be completely universal. Even for the same individual, the location of acupoints may change at different times and in different postures. 2. Skin condition: Skin color, texture, gloss and other characteristics may affect the effect of recognition technology. For example, skin that is too dry or greasy, scars or tattoos, etc., may interfere with the recognition process. III. Environmental interference: 1. Lighting conditions: The intensity and direction of light may affect the quality of the image and the recognition effect. Too strong or too weak light may cause the image to be blurred or distorted, thereby affecting the accuracy of acupoint recognition. 2. Clothing cover: In daily life, people usually wear clothes to cover their backs. The material, thickness, color, etc. of the clothes may affect the effect of the recognition technology. In particular, tight or heavy clothes may completely cover the acupoints, resulting in failure to recognize. IV. Operational complexity: 1. Equipment requirements: Human back contour and acupoint recognition technology usually relies on high-precision cameras, sensors and other equipment. These devices may be expensive and require professional operation and maintenance. 2. Data processing: The large amount of data generated during the recognition process needs to be processed and analyzed, which may require powerful computing power and professional software support. For non-professionals, this may be a huge challenge. V. Privacy and ethical issues: 1. Privacy leakage: When using relevant technologies for back contour and acupoint recognition, it may be necessary to collect and process personal biometric information. If this information is leaked or abused, it may pose a serious threat to personal privacy. 2. Ethical controversy: For certain specific groups of people, such as the elderly, children, and the disabled, the use of relevant technologies for back contour and acupoint recognition may involve ethical issues. For example, whether these groups should be subject to mandatory identification or monitoring, and how to ensure that their rights and interests are protected. Summary of the invention

[0005] The purpose of the embodiments of the present disclosure is to provide a method, device, computer-readable storage medium and computer program product for identifying acupuncture points on the back of the human body, so as to solve the above-mentioned problems existing in the prior art.

[0006] In order to achieve the above objectives, the technical solutions adopted by the embodiments of the present disclosure are as follows:

[0007] The present disclosure provides a method for identifying acupuncture points on the back of a human body, the method comprising:

[0008] Acquire a back image of a human body to be identified, wherein the Dazhui acupoint and the lumbar sacral vertebrae are preset with positioning icons in the back image;

[0009] Respectively identifying the positioning icons of the Dazhui acupoint and the lumbar sacral vertebra in the back image;

[0010] Determine the center points of two positioning icons in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoints and lumbar sacral vertebrae respectively;

[0011] According to the center points of the two positioning icons, the area between the Dazhui point and the lumbar sacral vertebrae is divided into multiple points.

[0012] Exemplarily, spinal points and bladder meridian points are marked based on the back image and the points obtained by averaging.

[0013] Exemplarily, the back image includes: a color image, and the positioning icons for respectively identifying the Dazhui acupoint and the lumbar sacral vertebra in the back image include:

[0014] Convert color images from BGR color space to HSV color space;

[0015] Perform principal component analysis on the pixel data in the HSV space and project the HSV color information onto the new coordinate system defined by PCA;

[0016] According to the preset threshold, the PCA projection result data is segmented and filtered to obtain a binary image, in which the positioning icon is the foreground and the rest is the background.

[0017] Exemplarily, determining the center point of each positioning icon in the back image includes:

[0018] Calculate the center point of each positioning icon in the binary image;

[0019] The center point of each positioning icon is mapped from the current space to the original color image to obtain the center point of each positioning icon in the back image.

[0020] Exemplarily, the positioning icon includes: a single color icon.

[0021] Exemplarily, the positioning icon includes: a green icon.

[0022] Exemplarily, the back image includes: a color image and a depth image. After dividing the area between the Dazhui point and the lumbar sacral vertebrae into a corresponding plurality of points, the method further includes:

[0023] Aligning the depth image and the color image and extracting features to obtain back contour data of a human body;

[0024] Based on the width of the human body's waist and back contour, the points obtained by evenly dividing the points and the preset distance, the acupuncture points of the bladder meridian on both sides of the spine are obtained.

[0025] Exemplarily, the back image includes: a color image and a depth image. After dividing the area between the Dazhui point and the lumbar sacral vertebrae into a corresponding plurality of points, the method further includes:

[0026] Aligning the depth image and the color image and extracting features to obtain back contour data of a human body;

[0027] According to the width of the waist and back contour of the human body, it is judged whether the Dazhui acupoint and the lumbar sacral vertebra positioning icons are accurate.

[0028] Exemplarily, judging whether the Dazhui acupoint and the lumbar sacral vertebra positioning icons are accurate according to the width of the waist and back contour of the human body, respectively, includes:

[0029] If the distance difference between the center of the Dazhui acupoint location icon and the two side edges in the width direction of the human back is less than a first threshold, the Dazhui acupoint location icon is accurate;

[0030] If the distance difference between the center of the lumbar vertebrae positioning icon and the two side edges in the width direction of the human back is less than a second threshold, the lumbar vertebrae positioning icon is accurate.

[0031] Exemplarily, after obtaining a plurality of points by evenly dividing, the method further comprises: determining the spinal acupoint and bladder meridian acupoint corresponding to each point according to a preset massage control mode of the acupoint.

[0032] Another aspect of the present disclosure provides a device for identifying acupuncture points on the back of a human body, the device comprising:

[0033] An acquisition module is used to acquire a back image of a human body to be identified, wherein the Dazhui acupoint and the lumbar sacral vertebrae in the back image are preset with positioning icons;

[0034] An image processing module, used to respectively identify the positioning icons of the Dazhui acupoint and the lumbar sacral vertebrae in the back image;

[0035] A center point determination module, used to determine the center points of two positioning icons in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoints and lumbar sacral vertebrae respectively;

[0036] The acupoint determination module is used to divide the area between the Dazhui acupoint and the lumbar sacral vertebrae into multiple points according to the center points of the two positioning icons.

[0037] On the other hand, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the steps of the above-mentioned method when executed by a processor.

[0038] Another aspect of the embodiments of the present disclosure provides a computer program product, including a computer program, which implements the steps of the method described above when executed by a processor.

[0039] The beneficial effects of the embodiments of the present disclosure are:

[0040] The method for identifying acupuncture points on the back of the human body in the disclosed embodiment can accurately identify other acupuncture points between the Dazhui point and the lumbar sacral vertebrae based on the preset positioning icons of the Dazhui point and the lumbar sacral vertebrae in the image of the human back, and does not require the upper body of the human body to be exposed, thereby protecting the privacy of the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of a method for identifying acupuncture points on the back of a human body provided by an embodiment of the present disclosure;

[0042] Figure 2 An image of a human back including a positioning icon obtained by using a method for identifying acupuncture points on the human back provided by an embodiment of the present disclosure;

[0043] Figure 3 The corresponding acupoints are obtained by using a method for identifying acupoints on the back of a human body provided by an embodiment of the present disclosure;

[0044] Figure 4 A schematic diagram of the structure of a device for identifying acupuncture points on the back of a human body provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the embodiment of the present disclosure more clear, the embodiment of the present disclosure is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation methods described here are only used to explain the embodiment of the present disclosure and are not used to limit the embodiment of the present disclosure.

[0046] The method for identifying acupoints on the back of the human body in the disclosed embodiment is applied to modern Chinese medicine intelligent mechanical arm massage, and the mechanical arm is provided with a massage head, and the massage head massages a certain area of ​​the human body during massage, that is, each massage head massages a certain massage area according to its massage control mode during massage. The spinal acupoints described in the disclosed embodiment are acupoints on the spine, and the bladder meridian acupoints are acupoints on the bladder meridian on both sides of the spine.

[0047] Embodiment 1:

[0048] like Figure 1 As shown, the present disclosure provides a method for identifying acupuncture points on the back of a human body, the method comprising:

[0049] Step S100, obtaining a back image of a human body to be identified, wherein positioning icons are preset for the Dazhui point and the lumbar sacral vertebra in the back image.

[0050] In step S100, the robotic arm is controlled to move to a specified position through pre-set spatial coordinates, and a depth camera is set on the robotic arm. The depth camera is used to collect the back image of the human body to be identified, mainly collecting the color image and depth image of the back of the human body; specifically, a gemini2 depth camera can be used to collect the back image of the human body. The gemini2 depth camera has two lenses, namely, an RGB lens and a depth lens. The RGB lens collects color pixel data of the back of the human body, and the depth lens collects depth image data of the back of the human body.

[0051] It should be noted that before the robotic arm obtains the image of the back of the human body to be identified, it is necessary to manually locate the two reference parts of the Dazhui point and the lumbar sacral vertebrae on the back spine of the human body to be identified, and affix pre-prepared positioning icons to the two parts of the Dazhui point and the lumbar sacral vertebrae respectively. The positioning icon here is a green icon, and the center of the positioning icon is best aligned with the corresponding part. The shape of the positioning icon can be circular, square, or triangular. The Dazhui point is located at the lower end of the human neck, and the lumbar sacral vertebrae is the part below the lumbar vertebrae and above the coccyx. To locate these two reference parts, the human body does not need to expose the upper body, which can well protect the user's privacy. The two reference parts of the Dazhui point and the lumbar sacral vertebrae can also be replaced with other convenient parts or acupoints according to actual conditions.

[0052] Step S200, respectively identifying the positioning icons of the Dazhui point and the lumbar sacral vertebrae in the back image.

[0053] Step S200 of the embodiment of the present disclosure needs to distinguish the outline of the positioning icon, that is, to distinguish the area of ​​the positioning icon from the background or other irrelevant parts in the human back image.

[0054] As a specific example of respectively identifying the positioning icons of the Dazhui acupoint and the lumbar sacral vertebrae in the back image, the back image includes: a color image, and the positioning icons of identifying the Dazhui acupoint and the lumbar sacral vertebrae in the back image include:

[0055] Step S210: Convert the color image from the BGR color space to the HSV color space to obtain HSV space pixels.

[0056] The disclosed embodiment uses the RGB lens of the gemini2 depth camera to obtain a color image of the back of the human body to be identified, and the identification of the positioning icon mainly uses the color image; step S210 converts the RGB of each pixel in the color image to the HSV space to obtain the single-point color value corresponding to each pixel in the HSV space, wherein the HSV space includes: H color channel, S color channel and V color channel. The single-point color value consists of hue H, saturation S and lightness / brightness V.

[0057] The HSV color space divides colors into three independent components: Hue (H): indicates the type of color, such as red, green, blue, etc. Saturation (S): indicates the purity or intensity of the color, that is, how much white is mixed into the color. Value (V): indicates the brightness of the color, that is, how much black is mixed into the color. HSV is more conducive to separating color information. For each pixel, the three components of hue H, saturation S, and value V are extracted, that is, through color space conversion, the HSV representation of all pixels in the color image is obtained.

[0058] After step S210, each color component (H, S, V) may be standardized, for example, by subtracting the mean and dividing by the standard deviation, to ensure that each dimension has the same scale.

[0059] Step S220: Perform principal component analysis on the pixel data in the HSV space, and project the HSV information onto a new coordinate system defined by PCA.

[0060] In the PCA process, step S220 of the embodiment of the present disclosure first calculates the covariance matrix or correlation matrix of the HSV pixel data, and solves its eigenvalues ​​and eigenvectors. The eigenvalues ​​reflect the size of the data variance explained by each principal component, while the eigenvectors define the directions of these principal components. Select important principal components: Sort by the size of the eigenvalues, and select the first few principal components with the largest eigenvalues ​​to retain the information that best represents the data structure.

[0061] Project the HSV pixel color values ​​into a new coordinate system consisting of the selected principal components. The result of this step is to convert high-dimensional data, such as the three-dimensional HSV space, into a low-dimensional representation, such as two or one dimension, while trying to keep the main features of the data unchanged. PCA will find the main directions of change of the data, that is, those directions that maximize the variance; in particular, the projection values ​​in the direction of the first principal component can usually best distinguish different color categories. For the green icon, the distribution of its pixels in the new coordinate system should be significantly different from other background pixels.

[0062] Before performing principal component analysis on the pixel data in the HSV space, the method further includes: data preparation and preprocessing;

[0063] Collect samples: First, you need to collect a series of RGB images containing positioning icons as a training set. These images should cover different lighting conditions, background changes, etc. as much as possible.

[0064] Convert color space: Convert all RGB images to HSV color space, because HSV is more conducive to separating color information. For each pixel, extract the three components of hue H, saturation S and lightness V.

[0065] The main purpose of collecting RGB images containing green icons is to:

[0066] Understanding data distribution: By collecting diverse samples, we can gain a more comprehensive understanding of how green icons behave under different conditions. This includes different lighting conditions, background complexity, changes in icon size and shape, etc.

[0067] Build a reference set: These samples form a reference set for subsequent PCA analysis. PCA will calculate the color features that can best distinguish the green icon from other backgrounds based on the color information in this reference set, such as H, S, and V values.

[0068] Improve robustness: Diverse samples can help the system adapt to various actual situations, thereby improving the accuracy and robustness of detection.

[0069] Step S230: segment and filter the PCA projection result data according to a preset threshold to obtain a binary image, in which the positioning icon is the foreground and the rest is the background.

[0070] It is also possible to reconstruct an image of the PCA projection result data according to the coefficients and principal components after PCA dimensionality reduction, and segment the reconstructed image to obtain a binary image, that is, to distinguish the positioning icon from the back image.

[0071] In step S230, a suitable threshold is selected to distinguish the positioning icon from other backgrounds according to the PCA projection result data. The preset threshold can be determined by histogram analysis or automatic threshold algorithm, such as Otsu's method.

[0072] After obtaining the binary image, the method may further include:

[0073] Step S240: Filter the binary image to remove noise or discontinuous small areas.

[0074] Due to the influence of actual shooting conditions, there may be noise or discontinuous small areas in the binary image. Gaussian filtering, median filtering and other techniques can be used to remove these small areas to achieve denoising.

[0075] In step S240, morphological operations may also be performed: an opening operation (erode first and then dilate) to eliminate small objects, or a closing operation (dilate first and then erode) to fill small holes to ensure that the recognition icon area is coherent and smooth.

[0076] Step S300: determining the center point of each positioning icon in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoint and lumbar sacral vertebrae, respectively.

[0077] After the positioning icon is identified in step S200, the center point of the positioning icon is calculated, and the center point of the positioning icon is matched with the corresponding Dazhui point and the lumbar sacral vertebra.

[0078] Step S310: Calculate the center point of each positioning icon in the binary image.

[0079] In step S310, the pixels of the positioning icon, i.e., the green icon, can be distributed on the coordinate system xoy, and the middle value can be taken to determine the position of the center point of the positioning icon. All independent areas in the image can also be identified through connected domain analysis, and the largest connected domain representing the green icon is marked. Then, the centroid calculation is performed: the average value of all pixel coordinates in the connected domain is calculated to obtain the position of the center point of the green icon.

[0080] Step S320: Map the center points of the positioning icons from the current space to the original color image to obtain the center points of the positioning icons in the back image.

[0081] After determining the center point of the positioning icon, it can be mapped from the HSV space or the binary image to the color image to obtain the center point of each positioning icon of the color image, such as Figure 2 shown.

[0082] Step S400: According to the center points of the two positioning icons, the area between the Dazhui point and the lumbar sacral vertebrae is evenly divided into a plurality of points to obtain a plurality of acupoints to be massaged.

[0083] In step S400, the embodiment of the present disclosure divides multiple points between the Dazhui point and the lumbar sacral vertebrae. Based on the knowledge of traditional Chinese medicine, according to the back contour map of the back image and the points obtained by the division, the acupoint data of the spinal points and the bladder meridian points obtained by the corresponding points are generated, so as to determine the specific position of each acupoint and realize precise massage. The spinal points are the acupoints on the spine, and the bladder meridian points are the acupoints on the bladder meridian on both sides of the spine. Figure 3As shown. The spinal acupoints and bladder meridian acupoints are calculated based on the center points of the two green icons. Although the line between the center points of the two green icons is evenly divided to obtain the approximate location of the corresponding acupoints, the treatment head has a certain area. Even if the points generated by the recognition are not within the acupoints, the acupoints are within the massage range of the treatment head, and the acupoints on the spine and the bladder meridians on both sides of the spine can be massaged. Specifically, 16 points can be evenly divided between the Dazhui acupoint and the lumbar sacral vertebrae. The robotic arm performs massage treatment based on the acupoint information.

[0084] Step S500: Mark the spinal points and bladder meridian points based on the back image and the points obtained by averaging.

[0085] Based on the knowledge of Chinese medicine anatomy, 16 points are evenly divided between the Dazhui point and the lumbar sacral vertebrae, and the spinal points and bladder meridian points are marked according to the back image and the points obtained by the even division. The number of points divided between the Dazhui point and the lumbar sacral vertebrae can also be set according to actual conditions.

[0086] If 16 points are evenly divided between the Dazhui point and the lumbar sacral vertebrae, and some points do not have corresponding spinal points, for points with corresponding spinal points, the points can be directly marked as corresponding spinal points. The spinal points from the Dazhui point to the lumbar sacral vertebrae include: Taodao point (corresponding to the first point), Shenzhu point (corresponding to the third point), Shendao point (corresponding to the fifth point), Lingtai point (corresponding to the sixth point), Zhiyang point (corresponding to the seventh point), Jinsu point (corresponding to the ninth point), Zhongshu point (corresponding to the tenth point), Jizhong point (corresponding to the eleventh point), Xuanshu point (corresponding to the thirteenth point), Mingmen point (corresponding to the fourteenth point) and Yaoyangguan point (corresponding to the sixteenth point) 11 acupoints. Some points on the Governor Vessel do not have corresponding acupoints, such as the eighth thoracic vertebra or lumbar vertebra; however, the bladder meridians on both sides of the Governor Vessel, which are parallel to the Governor Vessel, have acupoints corresponding to the equal-division points, from top to bottom: Dazhui, Fengmen, Feishu, Jueyinshu, Xinshu, Governor Vessel, Geshu, no acupoint (corresponding to the eighth thoracic vertebra), Ganshu, Gallbladdershu, Spleenshu, Stomachshu, Sanjiaoshu, Shenshu, Qihaishu, Dachangshu, Guanyuanshu (corresponding to the lumbar vertebra), 1.5 to 3 Cun is the bladder meridian. Find the bladder meridian points on both sides corresponding to each point of the Governor Vessel. Dazhui point to Dachangshu point correspond to the bladder meridian on both sides of the first to sixteenth points in sequence. The eighth point corresponds to the eighth thoracic vertebra, and there are no acupoints on both sides. For the bladder meridian points corresponding to the evenly divided points, do not directly mark the points as bladder meridian points, but take any point and two points on both sides of the Governor Vessel at a certain distance in the vertical direction as the bladder meridian points corresponding to the point, and mark the bladder meridian points corresponding to each point, such as Figure 3 As shown, then, focus on massaging the marked spinal points and bladder meridian points using corresponding massage methods.

[0087] The back image includes: a color image and a depth image. After the Dazhui point and the area between the lumbar vertebrae and the sacral vertebrae are equally divided into a plurality of corresponding points, the method further includes:

[0088] Step S510: aligning the depth image and the color image and extracting features to obtain back contour data of the human body.

[0089] The depth image data is aligned with the RGB image data to obtain the back contour data of the human body. Since the Gemini2 depth camera has two lenses (RGB lens and depth lens), the relative position between them is fixed, so the geometric relationship between the two can be established through calibration. The purpose of image alignment is to ensure that each pixel in the depth image can accurately correspond to the corresponding position in the RGB image. Intrinsic and extrinsic calibration: First, the two cameras need to be calibrated separately to obtain their respective intrinsic parameters (focal length, principal point, etc.) and relative position and rotation angle (external parameters). This step can be completed using a standard camera calibration plate and algorithm. Alignment: Using the above calibration parameters, the depth map is mapped to the RGB image coordinate system through a geometric transformation (such as an affine transformation or a perspective transformation), or vice versa.

[0090] After the image alignment is completed, the next step is to extract the contour information of the back of the human body from the depth image. Segmentation: Use depth threshold or other methods to separate the back area of ​​the human body from the depth image. Because the background and other objects usually have different depth values. Edge detection: Apply edge detection algorithms, such as the Canny edge detector, to find the contour boundary of the back of the human body and display it on the color image.

[0091] Step S520, according to the width of the waist and back contour of the human body, the points obtained by evenly dividing and the preset distance are obtained to obtain the acupuncture points of the bladder meridian on both sides of the spine.

[0092] The preset distance can be adjusted according to the width of the waist and back contour of the human body. If the waist and back contour of the human body is wider and greater than the preset threshold, the first preset distance on both sides of each point is the corresponding bladder meridian acupoint;

[0093] If the back contour of the human body is narrow, less than or equal to the preset threshold, the second preset distance on both sides of each point is the corresponding bladder meridian point. The width of the back of the human body can also be visually measured to adjust the preset distance.

[0094] The acupuncture points of the bladder meridian on both sides of the spine are obtained based on the proportional relationship between the width of the waist in the human body's waist and back contour and the distance from the Dazhui point to the lumbar sacral vertebra, and the points are evenly divided.

[0095] Step S520 combines the center point position of the green icon in the RGB image and the three-dimensional structure information provided by the depth image to further analyze the width of the waist and back of the human body. Width determination: Based on the human body contour in the depth image, the width of the back can be measured at different heights.

[0096] Horizontal positioning: According to the width of the back contour, the acupuncture points of the bladder meridian are located horizontally on both sides of each point according to the preset distance. According to the back contour obtained by the depth image, the treatment head can completely fit the back, making it easier to fit the treatment head to the back curve during massage.

[0097] After manually assisting in determining two reference positions, the massage arm or massage robot can automatically identify the user's back contour, spinal acupoints and bladder meridian acupoints based on the Dazhui point and the lumbar sacral vertebrae, thereby realizing the recognition of the human back contour and acupoint data without exposing the upper body. It greatly improves the user's privacy without affecting the accuracy of acupoint recognition, especially in some public places, and is very friendly to users who need to identify acupoints on the back of the human body.

[0098] like Figure 4 As shown, another aspect of the present disclosure provides a device for identifying acupuncture points on the back of a human body, the device comprising:

[0099] The acquisition module 100 is used to acquire a back image of a human body to be identified, wherein the Dazhui point and the lumbar sacral vertebrae in the back image are preset with positioning icons.

[0100] A depth camera is used to obtain the back image of the human body to be identified, and the back image of the human body includes: a color image and a depth image.

[0101] The image processing module 200 is used to respectively identify the positioning icons of the Dazhui point and the lumbar sacral vertebrae in the back image.

[0102] The image processing module 200 may use the method of step S200 to identify the location icons of the Dazhui point and the lumbar sacral vertebrae in the back image.

[0103] The center point determination module 300 is used to determine the center points of two positioning icons in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoints and lumbar sacral vertebrae respectively.

[0104] The center point determination module 300 can determine the center point of each positioning icon in the back image through step S300.

[0105] The acupoint determination module 400 is used to divide the area between the Dazhui acupoint and the lumbar sacral vertebrae into multiple points according to the center points of the two positioning icons.

[0106] The acupoint determination module 400 uses step S400 to obtain multiple equally divided points between the Dazhui acupoint and the lumbar sacral vertebrae, and can determine the spinal acupoints and bladder meridian acupoints corresponding to each point based on the massage control method of the acupoints. The spinal acupoints are the acupoints on the spine, and the bladder meridian acupoints are the acupoints on the bladder meridian on both sides of the spine.

[0107] The spinal points and bladder meridian points can also be marked based on the back image and the points obtained by averaging.

[0108] With manual assistance, the two reference acupuncture points on the human back spine, Dazhui and lumbar sacral, are determined and the positioning icons are pasted on them respectively. Then, the robotic arm is moved to the preset position. After the robotic arm automatically identifies the acupuncture points on the back of the human body, massage and other operations are performed.

[0109] The device further comprises: a bladder meridian acupoint determination module 500, which is used to align the depth image with the color image and extract features to obtain back contour data of a human body;

[0110] Based on the width of the human body's waist and back contour, the points obtained by evenly dividing the points, and the preset distance, the acupoints of the bladder meridian on both sides of the spine are obtained.

[0111] The bladder acupoint determination module 500 can use the method of step S510 and step S520 to obtain the bladder meridian acupoints on both sides of each equal-division point of the Governor Vessel.

[0112] The disclosed embodiment combines the principal component analysis algorithm with a depth camera, and uses green icons physically attached to the Dazhui point and lumbar sacral vertebrae on the back of the human body as an aid, so as to identify the relevant acupuncture points on the back spine of the human body without exposing the upper body, thereby avoiding privacy leakage.

[0113] Embodiment 2:

[0114] On the basis of the first embodiment, after determining the center points of the two positioning icons in the back image in step S300, it is necessary to determine whether the Dazhui acupoint and the lumbar sacral vertebra positioning icons are accurate. Specifically, the back image includes: a color image and a depth image. The method further includes:

[0115] Step S330: aligning the depth image and the color image and extracting features to obtain back contour data of the human body.

[0116] Step S340 can use the same method as the above step S510 to obtain the back contour data of the human body.

[0117] Step S350: judging whether the Dazhui acupoint and the lumbar sacral vertebra positioning icons are accurate according to the width of the waist and back contour of the human body.

[0118] Specifically include:

[0119] Step S351: If the distance difference between the center of the Dazhui acupoint positioning icon and the two side edges in the width direction of the human back is less than a first threshold, the Dazhui acupoint positioning icon is accurate; otherwise, it is inaccurate.

[0120] Step S352: If the distance difference between the center of the lumbar sacral vertebra positioning icon and the two side edges in the width direction of the human back is less than a second threshold, the lumbar sacral vertebra positioning icon is accurate; otherwise, it is inaccurate.

[0121] The first threshold and the second threshold can be set according to actual conditions. The first threshold can be 5 mm, and the second threshold can be 3 mm.

[0122] In the second embodiment, the spinal acupoints and bladder meridian acupoints corresponding to each point can also be determined according to the massage control method of the acupoints. In this embodiment, after the equal-dividing points on the Governor Vessel are determined in combination with the corresponding massage method, the spinal acupoints and bladder meridian acupoints corresponding to each point can be massaged in combination with the knowledge of Chinese medicine anatomy and according to the pre-set massage method. The corresponding spinal acupoints on the Governor Vessel basically correspond to the adjacent equal-dividing points. Some equal-dividing points do not have corresponding spinal acupoints. There are generally bladder meridian acupoints corresponding to the horizontal sides of the equal-dividing points. Therefore, after the equal-dividing points on the Governor Vessel are determined, when the pre-set massage method is used to massage each point, the spinal acupoints and bladder meridian acupoints corresponding to each point can be determined, and the corresponding massage can be performed. In combination with the first embodiment, the human waist and back contour is obtained according to the depth image and the color image. The height of each pixel of the human waist and back contour can be determined according to the depth image. If a special human back is encountered, the height of each average point needs to be determined. The height of each average point can be determined according to the average value of the height of each pixel in the first preset area where each average point is located, or the height of the pixel at each average point is the first height of the average point. The first preset area can be regarded as the massage area of ​​the corresponding spinal acupoint of the Governor Vessel; the massage area on both sides of each average point of the preset massage method is the second area. The second area can be regarded as the area of ​​the bladder meridian acupoints corresponding to the corresponding average point. According to the height of each pixel in the second area, the second height of the corresponding bladder meridian acupoint can be determined. If the difference between the first height and the second height is greater than the preset height, it means that the shape of the human back is relatively special. The massage method can be adjusted by increasing the massage intensity and / or expanding the massage area. Specifically, the range of the second area can be expanded to expand the massage range of the bladder meridian acupoints.

[0123] On the other hand, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the steps of the above-mentioned method when executed by a processor.

[0124] Another aspect of the embodiments of the present disclosure provides a computer program product, including a computer program, which implements the steps of the method described above when executed by a processor.

[0125] The above is only a preferred implementation of the embodiment of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the embodiment of the present disclosure. These improvements and modifications should also be considered as the protection scope of the embodiment of the present disclosure.

Claims

1. A method for identifying acupuncture points on the back of a human body, characterized in that: The method comprises: Acquire a back image of a human body to be identified, wherein the Dazhui acupoint and the lumbar sacral vertebrae are preset with positioning icons in the back image; Respectively identifying the positioning icons of the Dazhui acupoint and the lumbar sacral vertebra in the back image; Determine the center points of two positioning icons in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoints and lumbar sacral vertebrae respectively; According to the center points of the two positioning icons, the area between the Dazhui point and the lumbar sacral vertebrae is divided into multiple points.

2. The method according to claim 1, characterized in that The spinal points and bladder meridian points are marked based on the back image and the points obtained by averaging.

3. The method according to claim 1, characterized in that The back image includes: a color image, and the positioning icons for respectively identifying the Dazhui acupoint and the lumbar sacral vertebra in the back image include: Convert color images from BGR color space to HSV color space; Perform principal component analysis on the pixel data in the HSV space and project the HSV color information onto the new coordinate system defined by PCA; According to the preset threshold, the PCA projection result data is segmented and filtered to obtain a binary image, in which the positioning icon is the foreground and the rest is the background.

4. The method according to claim 3, characterized in that The determining the center point of each positioning icon in the back image includes: Calculate the center point of each positioning icon in the binary image; The center point of each positioning icon is mapped from the current space to the original color image to obtain the center point of each positioning icon in the back image.

5. The method according to any one of claims 1 to 4, characterized in that: The back image includes: a color image and a depth image. After the Dazhui point and the area between the lumbar vertebrae and the sacral vertebrae are equally divided into a plurality of corresponding points, the method further includes: Aligning the depth image and the color image and extracting features to obtain back contour data of a human body; Based on the width of the human body's waist and back contour, the points obtained by evenly dividing the points and the preset distance, the acupuncture points of the bladder meridian on both sides of the spine are obtained.

6. The method according to any one of claims 1 to 4, characterized in that: The back image includes: a color image and a depth image. After determining the center point of each positioning icon in the back image, the method further includes: Aligning the depth image and the color image and extracting features to obtain back contour data of a human body; According to the width of the waist and back contour of the human body, it is judged whether the Dazhui acupoint and the lumbar sacral vertebra positioning icons are accurate.

7. The method according to any one of claims 1 to 4, characterized in that: After the multiple points are evenly divided, the method further includes: determining the spinal acupoint and bladder meridian acupoint corresponding to each point according to the preset massage control mode of the acupoint.

8. A device for identifying acupuncture points on the back of a human body, characterized in that: The device comprises: An acquisition module is used to acquire a back image of a human body to be identified, wherein the Dazhui acupoint and the lumbar sacral vertebrae in the back image are preset with positioning icons; An image processing module, used to respectively identify the positioning icons of the Dazhui acupoint and the lumbar sacral vertebrae in the back image; A center point determination module, used to determine the center points of two positioning icons in the back image, wherein the center points of the two positioning icons correspond to the corresponding Dazhui acupoints and lumbar sacral vertebrae respectively; The acupoint determination module is used to divide the area between the Dazhui acupoint and the lumbar sacral vertebrae into multiple points according to the center points of the two positioning icons.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images