Acupuncture Location Method Based on Image Processing

Abstract

The present invention discloses an acupuncture point positioning method based on image processing, which relates to the technical field of traditional Chinese medicine. The present invention identifies feature points, acupoints, associated information and correction coefficients by constructing two neural network models for feature recognition of picture data; classifies by constructing a standard database and adopting a position tree structure; can quickly compare when the model inputs picture data, reduce the amount of computation and improve the accuracy of recognition; the present invention can directly mark the acupoint positions on the human body in cooperation with a laser or projection device, and doctors only need to perform acupuncture treatment according to the acupoint positions.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional Chinese medicine, and particularly to an acupuncture point positioning method based on image processing. Background Art

[0002] Acupuncture is the general term for acupuncture and moxibustion.

[0003] Acupuncture refers to inserting a needle (usually a filiform needle) into the patient's body at a certain angle under the guidance of traditional Chinese medicine theory, and using acupuncture techniques such as twirling and lifting-thrusting to stimulate specific parts of the human body to achieve the purpose of treating diseases. The insertion point is called the acupoint of the human body, abbreviated as the acupuncture point. According to the latest acupuncture and moxibustion textbooks, there are 361 regular acupuncture points in the human body; therefore, how to accurately judge the specific location of the acupuncture point is an essential skill in traditional Chinese medicine acupuncture; the bone measurement method is a common positioning method in traditional Chinese medicine, and this method can be used for positioning regardless of the patient's gender, age, height, or weight; in the process of learning acupuncture points, it generally relies on the doctor's experience and is assisted by a bone measurement ruler for positioning;

[0004] In the prior art, it mainly relies on the experience of traditional Chinese medicine and the bone measurement ruler for auxiliary positioning; the accuracy and precision of positioning are greatly affected by the personal experience of traditional Chinese medicine; with the development of big data and image recognition technology, computer image processing can quickly identify acupuncture points and can play an auxiliary role in the process of traditional Chinese medicine acupuncture; how to use big data technology to improve the accuracy of acupuncture point positioning has become an urgent technical problem to be solved. Summary of the Invention

[0005] In view of the problems existing in the above background, the present invention is proposed.

[0006] The problem to be solved by the present invention is how to accurately identify the position of acupuncture points from pictures using image recognition technology.

[0007] To solve the above technical problems, the present invention provides the following technical solutions:

[0008] In the first aspect of the present invention, an acupuncture point positioning method based on image processing is provided, including the following specific steps:

[0009] S1: Collect image data of acupuncture points in a public database;

[0010] S2: Preprocess the collected image data;

[0011] S3: Construct a first neural network model to perform feature recognition on the collected image information;

[0012] S4: Establish a database of feature points and classify them according to different parts;

[0013] S5: In the constructed database, identify and retain each joint point that has corresponding points in the bone measurement method.

[0014] S6: Based on the bone measurement method, with the located joint points as the starting or ending points, determine each preset line, and calculate the correction coefficient k corresponding to these lines.

[0015] S7: Use the data of different sizes and models collected in step S1 as the training set, and use the size data of the standard model in the bone measurement method as the validation set to train the second neural network model; after training, the second neural network model outputs the predicted correction coefficient k1 between the data of different sizes and models and the standard model.

[0016] S8: Collect the image information of the acupuncture site, identify the feature points through the first neural network model, and determine the predicted correction coefficient k1 through the second neural network model; classify the feature point database with different correction coefficients as the identification bits to construct a database with a tree structure.

[0017] S9: According to the predicted correction coefficient k1 of the person to be measured and the positions of each located joint point, determine the specific positions of each acupoint of the person to be measured according to the positioning rules of human acupoints.

[0018] Preferably, the image data collected in step S1 from the public database includes data of different parts, different sizes or different models, as well as the specific positions and physiological characteristics of each acupoint.

[0019] Preferably, in step S2, the collected data is subjected to data standardization processing, and image cleaning, image enhancement and segmentation processing steps are performed for subsequent feature recognition.

[0020] Preferably, in step S3, based on the cascaded convolutional neural network model and combined with the joint points in the bone measurement method, feature recognition is performed on the collected data, and the identified feature points are put into one-to-one correspondence with the joint points in the bone measurement method.

[0021] Preferably, in step S4, a data set with feature point coordinate data as elements is established for each different part.

[0022] Preferably, the correction coefficient in step S6 is calculated according to the following formula:

[0023] ;

[0024] Wherein, k is the correction coefficient; L a is the actually measured length between the preset lines; L sis the standard length of this line in the standard model; based on this, the line between feature point a and feature point b A has a standard length in the standard model of A 1 , then the line A has the following calculation formula:

[0025] ;

[0026] Calculating all the lines between different feature points can obtain all the feature position data within the image area;

[0027] Let the height be H , then calculate the height of each part according to the bone measurement method:

[0028] Head height: ;

[0029] Upper body length: ;

[0030] Leg length: ;

[0031] Chest circumference: ;

[0032] Waist circumference: ;

[0033] Hip circumference: ;

[0034] In the formula: H h is the head height; H u is the upper body length; H l is the leg length; C c is the chest circumference perimeter, K c is the chest circumference individual difference correction coefficient, with a value range of 1.5 - 2; C w is the waist circumference perimeter, K w is the waist circumference individual difference correction coefficient, with a value range of 1.0 - 1.5; C h is the hip circumference perimeter, K h is the hip circumference individual difference correction coefficient, with a value range of 1.5 - 2.0.

[0035] Preferably, the tree-structured database is retrieved by using a tree index; the database establishes identification codes according to the parent level, child level, and grandchild level;

[0036] Among them, the parent level is classified and coded in order according to the acupuncture area; including the head, face, chest, abdomen, waist, inner side of the calf, and inner side of the foot;

[0037] The child level is classified and coded in order according to the characteristic points in the corresponding area; the grandchild level is classified and coded in order according to the size of the prediction correction coefficient;

[0038] When a new image is collected, the region to which it belongs is first determined, and the database of the corresponding region is searched; then, the feature points are identified, and the prediction correction coefficients k1 in 2-3 lines are calculated in chronological order according to the identified feature points; the data under the same prediction correction coefficient k1 in the corresponding database is retrieved, and this is used as a verification set, and the subsequent identified feature points are compared using random numbers. If the comparison is successful, the results are quickly output based on the feature points and acupoint locations recorded in the database; if the comparison is unsuccessful, more feature points are identified until all feature points in the image area are identified, a new prediction correction coefficient is generated, and the new prediction correction coefficient is entered into the database.

[0039] Preferably, the following formula is used to calculate the prediction correction coefficients in 2-3 lines according to the characteristic points:

[0040] ;

[0041] In the formula, k 11 、k 12 、k 1n are the prediction correction coefficients in the 1st, 2nd, and nth lines respectively; n is the total number of predicted lines;

[0042] The calculated prediction correction coefficient k1 is traversed and searched with the data under several adjacent grandchildren codes in the database to improve the data comparison speed.

[0043] A second aspect of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned acupuncture positioning method based on image processing when executing the computer program.

[0044] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-mentioned acupuncture positioning method based on image processing.

[0045] The beneficial effects of the present invention are:

[0046] The present invention identifies the feature points, acupoints, associated information, and correction coefficients of image data by constructing two neural network models; classifies them by constructing a standard database and using a position tree structure; can quickly compare when the model inputs image data, reducing the amount of computation and improving the accuracy of identification; the present invention can directly mark the acupoint positions on the human body in cooperation with a laser or projection device, and doctors only need to perform acupuncture treatment according to the acupoint positions. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0048] Figure 1 It is a flowchart of an acupuncture location method based on image processing.

[0049] Figure 2 It is a schematic diagram of the positions of the Ruzhong acupoint and the Shanzhong acupoint. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings in the specification.

[0051] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0052] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that mutually excludes other embodiments.

[0053] Embodiment 1: As Figure 1 shown, the present invention provides an acupuncture location method based on image processing, including the following specific steps:

[0054] S1: Collect the image data of acupuncture acupoints in the public database;

[0055] In step S1, the image data of the acupoint database in the public platform is collected, including data of different parts, different sizes or different models, as well as the specific positions and physiological characteristics of each acupoint. For example, the image data of acupoint data maps of different parts can be collected from the acupoint database on the Chinese Medicine Resources Network (www.tcmdoc.cn);

[0056] S2: Preprocess the collected image data;

[0057] In step S2, the collected data is subjected to data standardization processing, and image cleaning, image enhancement and segmentation processing steps are performed for subsequent feature recognition;

[0058] S3: Construct a first neural network model to perform feature recognition on the collected image information;

[0059] In step S3, based on the cascaded convolutional neural network model and combined with the joint points in the bone measurement method, feature recognition is performed on the collected data, and the recognized feature points are put into one-to-one correspondence with the joint points in the bone measurement method. The cascaded convolutional neural network (Cascaded CNN) gradually and precisely locates the feature points through multiple cascaded CNN models; each cascaded CNN model further optimizes and adjusts the output of the previous model; the cascaded CNN model can gradually narrow the search range of the feature points and improve the positioning accuracy. At the same time, since each cascaded model is relatively lightweight, a fast running speed can be achieved while ensuring accuracy.

[0060] S4: Establish a database of feature points and classify them according to different parts;

[0061] In step S4, a data set with feature point coordinate data as elements is established for each different part respectively.

[0062] S5: In the constructed database, identify and retain each joint point that has a corresponding point in the bone measurement method; Table 1 below shows the common bone measurement table; Table 2 shows the table of positioning joint points in the bone measurement table;

[0063] Table 1 Common bone measurement table

[0064] Location Start and end points Discount Metrics illustrate head From the middle of the front hairline to the middle of the back hairline 12 inches Direct measurement Used to determine the longitudinal distance of acupoints on the head From the middle of the eyebrow to the front hairline 3 inches Direct measurement Used to determine the longitudinal distance between the front or back hairline and its head acupoints Between the hairlines of the two temples 9 inches Horizontal amount Used to determine the lateral distance of acupoints on the front of the head Between the two mastoid processes behind the ear 9 inches Horizontal amount Used to determine the lateral distance of acupoints at the back of the head Chest, abdomen and flank From the suprasternal fossa to the midpoint of the sternocostal junction 9 inches Direct measurement Used to determine the longitudinal distance of the Renmai acupoints on the chest From the midpoint of the sternocostal junction to the middle of the navel 8 inches Direct measurement Used to determine the longitudinal distance of acupoints in the upper abdomen From the middle of the umbilicus to the upper edge of the pubic symphysis 5 inches Direct measurement Used to determine the longitudinal distance of the lower abdominal acupoints Between the nipples 8 inches Horizontal amount Used to determine the lateral distance of chest and abdominal acupoints Back and waist From medial edge of scapula to posterior midline 3 inches Horizontal amount Used to determine the horizontal distance of the back and waist acupoints Upper limbs From the anterior and posterior axillary lines to the transverse lines of the elbow (at the level of the ulnar olecranon) 9 inches Direct measurement Used to determine the longitudinal distance of acupoints in the upper arm From the elbow crease (at the level of the olecranon) to the distal crease of the palmar (dorsal) side of the wrist 12 inches Direct measurement Used to determine the longitudinal distance of acupoints on the forearm Lower limbs From the upper edge of the pubic symphysis to the patellar base 18 inches Direct measurement Used to determine the longitudinal distance of acupoints on the thigh Patellar base to medial malleolus tip (below the medial tibial condyle) 13+15 inches Direct measurement Measure the distance to the lower part of the medial tibial condyle and the tip of the medial malleolus respectively Greater trochanter of femur to popliteal crease (level with patellar tip) 19 inches Direct measurement Used to determine the longitudinal distance of the acupoints on the anterolateral part of the thigh From gluteal groove to popliteal crease (level with patellar tip) 14 inches Direct measurement Used to determine the longitudinal distance of acupoints on the back of the thigh Popliteal crease (at the level of patellar tip) to lateral malleolus tip 16 inches Direct measurement Used to determine the longitudinal distance of the acupoints on the lateral side of the lower leg

[0065] Table 2 Table of positioning joint points in the bone measurement table

[0066] Positioning joints Starting point End point size illustrate From the middle of the front hairline to the middle of the back hairline Middle of front hairline Middle of the back hairline 12 Used to determine the longitudinal distance of acupuncture points on the head From the brow (Yintang) to the middle of the front hairline Between eyebrows (Yintang) Middle of front hairline 3 - From below the spinous process of the 7th cervical vertebra (Dazhui) to the middle of the posterior hairline Below the spinous process of the 7th cervical vertebra (Dazhui) Middle of the back hairline 3 - From the brow (Yintang) to the middle of the back hairline, below the spinous process of the 7th cervical vertebra (Dazhui) Between eyebrows (Yintang) The middle of the posterior hairline is below the spinous process of the 7th cervical vertebra (Dazhui) 18 - Between the two hair corners of the forehead Left frontal hair corner Right forehead hair corner 9 Used to determine the lateral distance of the acupuncture points on the front of the head Between the two mastoid processes (bone) behind the ear Left mastoid process Right mastoid 9 Used to determine the lateral distance of the acupuncture points at the back of the head From the suprasternal fossa (Tiantu) to the midpoint of the sternoxiphoid joint (Xiugu) Suprasternal fossa (Tiantu) Midpoint of sternoxiphoid joint (discussus) 9 Used to determine the longitudinal distance of the Ren Meridian acupoints on the chest From the midpoint of the sternocostal joint (the xiphoid process) to the middle of the umbilicus (the sacrum) Midpoint of sternoxiphoid joint (discussus) Umbilicus (Shenque) 8 Used to determine the longitudinal distance of acupuncture points in the upper abdomen From the middle of the umbilicus (Shenque) to the upper edge of the pubic symphysis (Qugu) Umbilicus (Shenque) Upper edge of pubic symphysis (pubic bone) 5 Used to determine the longitudinal distance of the acupuncture points in the lower abdomen Between the nipples Left nipple Right nipple 8 Used to determine the lateral distance of chest and abdominal meridian points From the axillary apex to the free end of the 11th rib (Zhangmen) Axillary apex Free end of 11th rib (Zhangmen) 12 Used to determine the lateral distance of the acupoints in the flank From the inner edge of the scapula (close to the lateral point of the spine) to the posterior midline Inner edge of scapula Posterior midline 3 For determining the transverse distance of acupoints on the back and waist From the acromial margin to the posterior midline Acromial margin Posterior midline 8 For determining the transverse distance of acupoints on the shoulder and back From the anterior and posterior axillary creases to the cubital crease (level with the cubital tip) Anterior and posterior axillary creases Cubital crease (level with the cubital tip) 9 For determining the longitudinal distance of acupoints on the arm From the cubital crease (level with the cubital tip) to the transverse crease on the palmar (dorsal) side of the wrist Cubital crease (level with the cubital tip) Transverse crease on the palmar (dorsal) side of the wrist 12 For determining the longitudinal distance of acupoints on the forearm From the upper border of the pubic symphysis to the upper border of the medial epicondyle of the femur Upper border of the pubic symphysis Upper border of the medial epicondyle of the femur 18 For determining the longitudinal distance of acupoints of the three yin meridians of the foot on the medial side of the lower limb From below the medial condyle of the tibia to the tip of the medial malleolus Below the medial condyle of the tibia Tip of the medial malleolus 13 - From the greater trochanter of the femur to the popliteal crease Greater trochanter of the femur Popliteal crease 19 - From the popliteal crease to the tip of the lateral malleolus Popliteal crease Tip of the lateral malleolus 16 -

[0067] S6: Based on the bone measurement method, taking the positioning joint points as the starting point or the ending point, determine each preset line and calculate the correction coefficients corresponding to these lines k ;

[0068] In step S6, the correction coefficient is calculated according to the following formula:

[0069] ;

[0070] where, k is the correction coefficient; L a is the actually measured length between the preset lines; L s is the standard length of this line in the standard model; based on this, the standard length of line A between feature point a and feature point b in the standard model is A 1 , then the calculation formula of line A is as follows:

[0071] ;

[0072] By calculating all the lines between different feature points, all the feature point data within the image area can be obtained;

[0073] Let the height be H , then calculate the height of each part according to the bone measurement method:

[0074] Head height: ;

[0075] Upper body length: ;

[0076] Leg length: ;

[0077] Chest circumference: ;

[0078] Waist circumference: ;

[0079] Hip circumference: ;

[0080] In the formula: H h is the head height; H u is the upper body length; H l is the leg length; C c is the chest circumference perimeter, K c is the chest circumference individual difference correction coefficient, with a value range of 1.5 - 2; C w is the waist circumference perimeter, K w is the waist circumference individual difference correction coefficient, with a value range of 1.0 - 1.5; Ch is the hip circumference, K h is the correction coefficient for individual differences in hip circumference, with a value of 1.5-2.0;

[0081] S7: using the data of different sizes collected in step S1 as a training set and the size data of the standard model in the bone measurement method as a verification set to train the second neural network model; after the training is completed, the second neural network model outputs a prediction correction coefficient k1 between the data of different sizes and the standard model;

[0082] S8: Collect image information of acupuncture sites, identify feature points through the first neural network model, and determine the prediction correction coefficient k1 through the second neural network model; classify the feature point database with different correction coefficients as identification points to build a tree-structured database;

[0083] The tree-structured database is searched by tree index; the database establishes identification codes according to the parent level, child level, and grandchild level;

[0084] Among them, the parent level is classified and coded in order according to the acupuncture area; including the head, face, chest, abdomen, waist, inner side of the calf, and inner side of the foot;

[0085] The child level is classified and coded in order according to the characteristic points in the corresponding area; the grandchild level is classified and coded in order according to the size of the prediction correction coefficient;

[0086] When a new image is collected, first determine the region to which it belongs, and search in the database of the corresponding region; then identify the feature points, and calculate the prediction correction coefficient k1 in 2-3 lines in order of the time of the identified feature points; retrieve the data under the same prediction correction coefficient k1 in the corresponding database, and use it as a verification set to compare the subsequent identified feature points in a random number manner. If the comparison is successful, the result is quickly output with the feature points and acupoint positions recorded in the database; if the comparison is unsuccessful, continue to identify more feature points until all feature points in the image area are identified, generate new prediction correction coefficients, and enter the new prediction correction coefficients into the database;

[0087] The following formula is used to calculate the prediction correction coefficient in 2-3 lines based on the characteristic points:

[0088] ;

[0089] In the formula, k 11 、k 12 、k 1nThey are the prediction correction coefficients in the 1st, 2nd, and nth lines respectively; n is the total number of prediction lines.

[0090] The calculated prediction correction coefficient k1 is retrieved by traversing the data under several adjacent sub - level codes in the database to improve the data comparison speed.

[0091] S9: According to the prediction correction coefficient k1 of the person to be measured and the positions of each positioning joint point, and in accordance with the positioning rules of human acupoints, determine the specific positions of each acupoint of the person to be measured.

[0092] The following uses a specific example to introduce the application process of this solution in detail.

[0093] Specifically, in this embodiment, take the chest acupoints as an example, as Figure 2 shown.

[0094] First, use the image acquisition module to collect a chest picture of the patient, and use the trained first neural network model to perform feature recognition on this picture. For a chest picture, the most easily determined feature point is the nipple (Ruzhong acupoint). According to the bone - measurement method, the Ruzhong acupoint is located in the center of the nipple, and the Tanzhong acupoint is located at the mid - point between the two nipples. The Tanzhong acupoint is four cun away from the Ruzhong acupoint. Based on this, by comparing the length of the preset line between the two Ruzhong acupoints with the standard data in the standard model, the correction coefficient can be obtained. For example, if the actual distance measured between the two Ruzhong acupoints is 6 cun, then compare it with the 4 cun in the standard model, and the prediction correction coefficient can be calculated using the formula. Then, taking the Ruzhong acupoint as the feature point, calculate the distances of other acupoints in the picture area and mark them in the figure. With the cooperation of a laser or projection device, the acupoint positions can be directly marked on the human body, and the doctor only needs to perform acupuncture treatment according to the acupoint positions.

[0095] In the database with a tree structure established in this embodiment, it is convenient to find the corresponding data, and the existing historical data can be quickly compared, thereby improving the recognition speed and accuracy. Specifically, first, different identification codes are set for each group of data in the database. The identification codes are arranged in the sequence of parent, child, and grandchild. For example, an 8-digit number is used as the identification code. The first two digits are the parent code, the middle three digits are the child code, and the last three digits are the grandchild code. Then the identification code 03023012 means that the data is stored in the area of parent level 03, child level 023, and grandchild level 012. Among them, the parent level is classified and coded in sequence according to the acupuncture areas. For example, the head is 01, the face is 02, the chest is 03, the abdomen is 04, the waist is 05, the inner side of the calf is 06, the inner side of the foot is 07, and so on. The child level is classified and coded in sequence according to the characteristic points in the corresponding area. For example, in the chest data, the characteristic point number of the Ruzhong acupoint is 023. The grandchild level is classified and coded in sequence according to the size of the prediction correction coefficient. Therefore, the above identification code 03023012 should retrieve the data segment with the correction coefficient number 012 of the Ruzhong acupoint as the characteristic point in the chest data.

[0096] When the characteristic points in the picture are recognized, they are sorted in chronological order. Taking the characteristic points a, b, c, d as an example, a is the first recognized characteristic point, and d is the last recognized characteristic point. When using this solution, first determine the area and characteristic points where the characteristic points are located, and then the parent and child data positions can be determined in the database. Then establish a line between a and b, calculate the correction coefficient, and obtain the first correction coefficient. Then establish lines between a and c, and b and c, calculate the correction coefficient, and obtain the second correction coefficient. When calculating the prediction correction coefficients in 2 - 3 lines according to the characteristic points, the following formula is used:

[0097] ;

[0098] In the formula, k 11 、k 12 、k 1n are the prediction correction coefficients in the 1st, 2nd, and nth lines respectively; n is the total number of prediction lines.

[0099] Use the prediction correction coefficient to retrieve the corresponding and the nearby 2 data segments. Use the data in the data segments as verification data for verification.

[0100] After that, establish circuits among ad, bd, and cd, calculate the correction factor, and obtain the third correction factor. Compare the third correction factor with the data segment in the database. If the similarity is greater than 99%, it is considered that the comparison is successful, and directly retrieve other feature points and acupoints within this data segment in the database for direct output, greatly reducing the amount of calculation. When the similarity of the correction factor is low, continue to find the next feature point and repeat the above steps until the comparison with the data segment is successful. If, after traversing all the feature points within the image area, the comparison is still not successful, then output the actual operation result of the model, save this data, and save the recognition code in a separate data segment in the database according to the above encoding steps.

[0101] Embodiment 2: The present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the acupuncture point positioning method based on image processing in Embodiment 1.

[0102] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of this computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0103] Embodiment 3: The present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-mentioned acupuncture point positioning method based on image processing.

[0104] The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM for short), electrically erasable programmable read-only memory (EEPROM for short), erasable programmable read-only memory (EPROM for short), programmable read-only memory (PROM for short), read-only memory (ROM for short), magnetic memory, flash memory, magnetic disk or optical disc.

[0105] In summary, the present invention uses the data in the existing public database and the bone measurement method to establish a standard model; analyzes and processes different types of acupoint models with different sizes and different parts on the market to construct a first neural network model and a second neural network model; the first neural network model is used to perform feature recognition on picture data, identify feature points, acupoints and related information; the second neural network model is used to output a correction coefficient; a standard database is constructed and classified using a position tree structure; when the model inputs picture data, it can be quickly compared, reducing the amount of computation and improving the accuracy of recognition; the present invention can directly mark the acupoint positions on the human body in cooperation with a laser or projection device, and doctors only need to perform acupuncture treatment according to the acupoint positions.

[0106] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An acupuncture positioning method based on image processing, characterized in that: The specific steps include: S1: Collect image data of acupuncture points in public databases; S2: preprocessing the collected image data; S3: constructing a first neural network model to perform feature recognition on the collected image information; S4: Establish a database of feature points and classify them according to different parts; S5: In the constructed database, identify and retain each joint point that has a corresponding point in the bone measurement method; S6: Based on the bone measurement method, the positioning joint point is used as the starting point or the end point to determine each preset route, and the correction coefficient k corresponding to these routes is calculated; S7: using the data of different sizes collected in step S1 as a training set and using the size data of the standard model in the bone measurement method as a verification set to train the second neural network model; After the training is completed, the second neural network model outputs the prediction correction coefficient k1 between the data of different size models and the standard model; S8: Collect image information of acupuncture sites, identify feature points through the first neural network model, and determine the prediction correction coefficient k1 through the second neural network model; classify the feature point database with different correction coefficients as identification points to build a tree-structured database; S9: According to the predicted correction coefficient k1 of the person to be tested and the positions of each positioning joint point, the specific position of each acupoint of the person to be tested is determined according to the positioning rules of the acupoints of the human body; In step S3, feature recognition is performed on the collected data based on the cascade convolutional neural network model combined with the joint points in the bone measurement method, and the identified feature points are matched one by one with the joint points in the bone measurement method; In step S6, the correction coefficient is calculated as follows: in, k is the correction factor; L a It is the length actually measured between the preset lines; L s is the standard length of the line in the standard model; based on this, the line between feature point a and feature point b A The standard length in the standard model is A 1 , then the line A The calculation formula is as follows: By calculating all the lines between different feature points, the data of all feature points in the image area can be obtained; Assume height H , then calculate the height of each part according to the bone measurement method: Head height: Upper body length: Leg length: chest circumference: waistline: Hips: Where: H h is the head height; H u is the length of the upper body; H l is the leg length; C c is the chest circumference, K c is the correction coefficient for individual differences in chest circumference, with a value of 1.5-2; C w is the waist circumference, K w is the correction coefficient for individual differences in waist circumference, with a value of 1.0-1.5; C h is the hip circumference, K h is the correction coefficient for individual differences in hip circumference, with a value of 1.5-2.0; The tree-structured database is searched by tree index; the database establishes identification codes according to the parent level, child level, and grandchild level; Among them, the parent level is classified and coded in order according to the acupuncture area; including the head, face, chest, abdomen, waist, inner side of the calf, and inner side of the foot; The child level is classified and coded in order according to the characteristic points in the corresponding area; the grandchild level is classified and coded in order according to the size of the prediction correction coefficient; When a new image is collected, the region to which it belongs is first determined, and the corresponding region database is searched; then the feature points are identified, and the prediction correction coefficients k1 in 2-3 lines are calculated in the order of the time of the identified feature points; the prediction correction coefficients k1 identical to the corresponding database are searched. The data under is used as the verification set, and the subsequent identified feature points are compared with random numbers. If the comparison is successful, the results are quickly output based on the feature points and acupoint positions recorded in the database; if the comparison is unsuccessful, more feature points are identified until all feature points in the image area are identified, a new prediction correction coefficient is generated, and the new prediction correction coefficient is entered into the database.

2. The acupuncture positioning method based on image processing as claimed in claim 1, characterized in that: The image data collected in the public database in step S1 include data of different parts, different sizes or different models, as well as the specific location and physiological characteristics of each acupoint.

3. The acupuncture positioning method based on image processing as claimed in claim 1, characterized in that: In step S2, the collected data is subjected to data standardization, and image cleaning, image enhancement and segmentation processing steps are performed for subsequent feature recognition.

4. The acupuncture positioning method based on image processing as claimed in claim 1, characterized in that: In step S4, a data set with feature point coordinate data as elements is established for each different part.

5. The acupuncture positioning method based on image processing as claimed in claim 1, characterized in that: The following formula is used to calculate the prediction correction coefficient in 2-3 lines based on the characteristic points: In the formula, k 11 、k 12 、k 1n are the prediction correction coefficients in the 1st, 2nd, and nth lines respectively; n is the total number of predicted lines; The calculated prediction correction coefficient k1 is traversed and searched with the data under several adjacent grandchildren codes in the database to improve the data comparison speed.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the acupuncture positioning method based on image processing described in any one of claims 1 to 5 are implemented.

7. 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 acupuncture positioning method based on image processing described in any one of claims 1 to 5 are implemented.

Images