Acupuncture Point Location Method and System Based on Image Enhancement

Abstract

The present invention discloses a method and system for acupoint location based on image enhancement, belonging to the technical field of image processing. The method includes: human acupoint image acquisition, human acupoint image enhancement, constructing an acupoint recognition model, and acupoint location. In this solution, through the parameter optimization of the gamma value and fractional order, according to the difference between the oscillation coefficient and the individual position, and introducing random perturbation for update, continuously optimize the parameter combination in combination with the evolution rate, and use the optimal parameter combination to perform gamma transformation and fractional order enhancement, so as to improve the image quality; the activation recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map, combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then performs weighting through global average pooling and soft attention mechanism, significantly improving the accuracy and stability of acupoint recognition.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and specifically refers to an acupoint positioning method and system based on image enhancement. Background Art

[0002] The acupoint positioning method is a method that uses image processing and deep learning technologies to analyze and locate the images of human acupoints, extract the characteristic information in the human acupoint images, automatically identify and mark the acupoints, reduce manual intervention, and improve the accuracy and automation level of positioning. However, in the existing acupoint positioning methods, there are problems such as poor quality of human acupoint images, unclear display of acupoint-related features, difficulty in accurately identifying and analyzing acupoint features, resulting in a large misrecognition rate of acupoint positioning; in the existing acupoint positioning methods, the features of human acupoint images are multi-scale and have complex local details, and the features of different scales often appear in different ways in the image. It is difficult for the existing methods to accurately extract from the details and make effective feature fusion at different scales, resulting in difficulties in accurately positioning the acupoint positions. Summary of the Invention

[0003] In view of the above situation, to overcome the defects of the prior art, the present invention provides an acupoint positioning method and system based on image enhancement. Aiming at the problems in the existing acupoint positioning methods, such as poor quality of human acupoint images, unclear display of acupoint-related features, difficult to accurately identify and analyze acupoint features, resulting in a large misidentification rate of acupoint positioning, this solution optimizes the parameters of the gamma value and fractional order. According to the difference between the oscillation coefficient and the individual position, and introducing random perturbation, the individual position is updated. The quantity is continuously updated according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, the gamma function and the fractional order differential algorithm are used to perform gamma transformation and fractional order enhancement on the human acupoint image, obtaining an enhanced human acupoint image, which more clearly displays acupoint-related features, provides reliable input for subsequent acupoint positioning, is conducive to the identification and analysis of acupoint features, and thus improves the accuracy of acupoint positioning. Aiming at the problems in the existing acupoint positioning methods, such as the features of human acupoint images having multi-scale and complex local details, and the features at different scales often being represented in different ways in the image, it is difficult for the existing methods to accurately extract details and make effective feature fusion at different scales, resulting in difficulties in accurately positioning the acupoint position. This solution designs an intelligent feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then weights them through global average pooling and a soft attention mechanism to obtain a feature fusion map, reducing redundant information and optimizing the fusion of multi-scale features, effectively extracting subtle and key acupoint features from complex human acupoint images, and significantly improving the accuracy and stability of acupoint recognition.

[0004] The technical solution adopted by the present invention is as follows: The acupoint positioning method based on image enhancement provided by the present invention includes the following steps:

[0005] Step S1: Acquisition of human acupoint images;

[0006] Step S2: Enhancement of human acupoint images;

[0007] Step S3: Construction of an acupoint recognition model;

[0008] Step S4: Acupoint positioning.

[0009] Further, in step S1, the acquisition of human acupoint images is to acquire human acupoint images marked with acupoints.

[0010] Further, in step S2, the enhancement of the human acupoint image optimizes the gamma value γ of the gamma function and the fractional order α of the fractional differential algorithm to find the optimal parameter combination (γ, α). Based on the optimal parameter combination (γ, α), the gamma function and the fractional differential algorithm are used to perform gamma transformation and fractional order enhancement on the human acupoint image to obtain the enhanced human acupoint image, and a training data set and a test data set are constructed based on the enhanced human acupoint image. The parameter optimization specifically includes the following steps:

[0011] Step S21: Initial individual positions; establish a parameter search space for the gamma value γ of the gamma function and the fractional order α of the fractional differential algorithm, and preset the number of initial individual positions H0, the minimum number of individual positions H min , the maximum number of searches T, and the fitness threshold f th . Randomly initialize H 0 individual positions within the parameter search space. Each individual position represents a set of parameter combinations. The information entropy, average gradient, and variance of the enhanced human acupoint image obtained based on the parameter combinations are weighted and combined as the fitness value corresponding to the individual position;

[0012] Step S22: Update individual positions; update each individual position according to the oscillation coefficient and the difference between individual positions, and introduce random perturbations. The formula used is as follows:

[0013] ;

[0014] ;

[0015] In the formula, and are the h-th individual positions at the (t + 1)-th and t-th searches respectively, Q t is the oscillation coefficient at the t-th search, h is the individual position index, t is the search number index, is a randomly selected individual position at the t-th search, r1 and r2 are two non-interfering random numbers, sin(·) and round(·) are the sine function and the rounding function respectively, and randn is a random number subject to a normal distribution;

[0016] Step S23: Search number detection; if the search number , then go to step S24; otherwise, go to step S25;

[0017] Step S24: Update quantity; includes the following steps:

[0018] Step S241: Calculate the evolution rate; the formula used is as follows:

[0019] ;

[0020] Wherein, L t+1 is the evolution rate during the (t + 1)-th search, and are the global optimal positions during the o-th and (o - 1)-th searches respectively, and are respectively and 's fitness values, o is the search times index, is the ceiling symbol, is the smoothing term;

[0021] Step S242: Evolution rate detection; If the evolution rate L t+1 is less than or equal to 10 -6 , then go to Step S243; Otherwise, go to Step S25;

[0022] Step S243: Update the number of individual positions; Calculate the new number of individual positions H t+1 based on the fitness value of the global optimal position. If H t+1 is less than or equal to H t , then sort all the individual positions in ascending order according to the fitness value, retain the first H t+1 individual positions, and discard the remaining individual positions; Otherwise, retain the original H t individual positions, and randomly initialize individual positions again within the parameter search space; The formula used is as follows:

[0023] ;

[0024] Wherein, H t , H t+1 and H 0.5T are the numbers of individual positions during the t-th, (t + 1)-th and 0.5T-th searches respectively, is the global optimal position during the (t + 1)-th search, is 's fitness value, is the minimum fitness value since the (t + 1)-th search;

[0025] Step S25: Determine the optimal parameter combination; Update the fitness values of the individual positions. When there exists a fitness value of the global optimal position less than the fitness threshold f th , then the parameter combination represented by the global optimal position is the optimal parameter combination (γ, α), and the parameter optimization is completed; Otherwise, if the maximum search times is reached, return to Step S21 to re-initialize the individual positions; Otherwise, increment the search times by 1 and return to Step S22 to continue the search.

[0026] Further, in step S3, the acupoint recognition model is constructed using the YOLOv8 network architecture; the YOLOv8 network architecture includes a backbone network, a neck network, and a detection head; the backbone network is responsible for extracting feature maps of different scales from the enhanced images of human acupoints in the training data set through convolutional operations; the neck network is responsible for further fusing and processing the multi-scale feature maps extracted by the backbone network; the detection head is responsible for acupoint recognition based on the feature maps output by the neck network; in the construction of the acupoint recognition model, a wisdom feature convolution is designed to replace the standard convolution in the backbone network, and the wisdom feature convolution consists of an activation recombination unit and a dual-type convolution optimization unit, which specifically includes the following steps:

[0027] Step S31: Activation recombination unit; includes the following steps:

[0028] Step S311: Group normalization; perform group normalization on the input feature map I;

[0029] Step S312: Calculate weights; normalize the scaling factors of each channel to obtain the weights corresponding to each channel;

[0030] Step S313: Channel gating activation; map the weighted feature map to the interval (0, 1) through the Sigmoid function, and preset the threshold W th for gating processing. For channels greater than the threshold W th , set their weights to 1 to obtain the activation weight W1. For channels less than the threshold W th , set their weights to 0 to obtain the inhibition weight W2; the formula used is as follows:

[0031] ;

[0032] In the formula, Gate(·) and Sigmoid(·) are the gating function and the activation function respectively, W a is the weight, a is the index, W1 and W2 are the activation weight and the inhibition weight respectively, W1 = 1, W2 = 0, is the weight set, is the feature map after group normalization processing;

[0033] Step S314: Split the input feature map; multiply I by W1 and W2 respectively, so as to split I into the feature map activation component and the feature map inhibition component ;

[0034] Step S315: Split and recombine; for acupoint recognition, diverse features can be extracted from different channels and details, and the interaction between different features can be improved through splitting and recombination, so as to capture the complex morphological information of acupoints; and are evenly split along the channels respectively, and two sub-feature maps of equal size are evenly split from , and two sub-feature maps of equal size are evenly split from and . Then, the cross combination is performed on the evenly split sub-feature maps to obtain two new feature maps I and I . Then, I and I ω1 and I ω2 are connected to obtain the composite feature map I ω1 and I ω2 ; ω ;

[0035] Step S32: Dual-type convolution optimization unit; includes the following steps:

[0036] Step S321: Split the composite feature map; Split I ω according to and channels to obtain the core feature map I upper and the detail feature map I below . Then, apply 1×1 convolution to compress the channel numbers of I upper and I below respectively, so as to obtain the core feature map after convolution compression and the detail feature map ; where is the control parameter, and U is the number of channels in I;

[0037] Step S322: Dual-type convolution combination; For , first perform grouped convolution and pointwise convolution respectively, and then add and to obtain the core convolution feature map O1; For , first perform pointwise convolution , and then connect and to obtain the detail convolution feature map O2;

[0038] Step S323: Feature reconstruction and optimization; Use global average pooling to process O1 and O2 respectively to obtain the core global feature B1 and the detail global feature B2, and then generate the core weight factor and the detail weight factor through the soft attention mechanism. Then, use and to perform weighted combination on O1 and O2 respectively to obtain the feature fusion map O.

[0039] Further, in step S4, the acupoint positioning is to collect the human acupoint image to be positioned and perform image enhancement to obtain the enhanced human acupoint image to be positioned, then input the enhanced human acupoint image to be positioned into the acupoint recognition model for processing, identify the area where the acupoint is located, and perform annotation, and output the annotated human acupoint image to complete the acupoint positioning.

[0040] The acupoint positioning system based on image enhancement provided by the present invention includes a human acupoint image acquisition module, a human acupoint image enhancement module, a module for constructing an acupoint recognition model, and an acupoint positioning module;

[0041] The human acupoint image acquisition module acquires the human acupoint image with acupoint annotation and sends the data to the human acupoint image enhancement module;

[0042] The human acupoint image enhancement module optimizes the parameters of the gamma value and the fractional order, updates the individual position according to the difference between the oscillation coefficient and the individual position, and introduces random perturbation, continuously updates the quantity according to the evolution rate, finds the optimal parameter combination, and performs gamma transformation and fractional order enhancement on the human acupoint image based on the optimal parameter combination using the gamma function and the fractional order differential algorithm to obtain the enhanced human acupoint image, and sends the data to the module for constructing the acupoint recognition model;

[0043] The module for constructing the acupoint recognition model designs a wisdom feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then performs weighting through global average pooling and a soft attention mechanism to obtain a feature fusion map, and sends the data to the acupoint positioning module;

[0044] The acupoint positioning module collects the human acupoint image to be positioned and performs image enhancement, then inputs it into the acupoint recognition model for processing, identifies the area where the acupoint is located, and performs annotation, and outputs the annotated human acupoint image.

[0045] The beneficial effects achieved by the present invention using the above solution are as follows:

[0046] (1) Aiming at the problem that in the existing acupoint location methods, the quality of human acupoint images is poor, the acupoint-related features are not clearly displayed, it is difficult to accurately identify and analyze acupoint features, resulting in a large misidentification rate of acupoint location. This solution optimizes the parameters of the gamma value and fractional order, and based on the difference between the oscillation coefficient and the individual position, introduces random perturbation to update the individual position. Continuously update the quantity according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, use the gamma function and fractional order differential algorithm to perform gamma transformation and fractional order enhancement on the human acupoint image to obtain an enhanced human acupoint image. Then construct a training data set and a test data set, improve the quality of the human acupoint image, more clearly display the acupoint-related features, provide reliable input for subsequent acupoint location, facilitate the identification and analysis of acupoint features, reduce the misidentification rate, and thus improve the accuracy of acupoint location.

[0047] (2) Aiming at the problem that in the existing acupoint location methods, the features of human acupoint images are multi-scale and have complex local details, and the features at different scales often appear in different ways in the image. It is difficult for existing methods to accurately extract details and make effective feature fusion at different scales, resulting in difficulties in accurately locating acupoint positions. This solution designs an intelligent feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then weights them through global average pooling and a soft attention mechanism to obtain a feature fusion map, realizing the effective extraction of detail and overall information. Thus, it improves the perception ability of complex acupoint shapes, reduces redundant information and optimizes the fusion of multi-scale features, ensures the accurate capture of features at different scales, effectively extracts subtle and key acupoint features from complex human acupoint images, and significantly improves the accuracy and stability of acupoint recognition. Description of the Drawings

[0048] Figure 1 It is a schematic flow chart of the acupoint location method based on image enhancement provided by the present invention;

[0049] Figure 2 It is a schematic diagram of the acupoint location system based on image enhancement provided by the present invention;

[0050] Figure 3 It is a schematic flow chart of step S2;

[0051] Figure 4 It is a schematic flow chart of step S3.

[0052] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. Detailed implementation manners

[0053] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0054] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention.

[0055] Embodiment 1. Refer to Figure 1 , the acupoint location method based on image enhancement provided by the present invention includes the following steps:

[0056] Step S1: Acquisition of human acupoint images; acquisition of human acupoint images marked with acupoints.

[0057] Step S2: Enhancement of human acupoint images; optimization of parameters for gamma value and fractional order, updating of individual positions according to the difference between the oscillation coefficient and the individual position, and introduction of random perturbations. Continuously update the quantity according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, use the gamma function and fractional order differential algorithm to perform gamma transformation and fractional order enhancement on the human acupoint image to obtain the enhanced human acupoint image.

[0058] Step S3: Construction of an acupoint recognition model; design of intelligent feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain the core convolution feature map and the detailed convolution feature map, and then performs weighting through global average pooling and soft attention mechanism to obtain the feature fusion map.

[0059] Step S4: Acupoint location; acquisition of the human acupoint image to be located and image enhancement, and then input it into the acupoint recognition model for processing, identify the area where the acupoint is located, and perform annotation, and output the annotated human acupoint image.

[0060] Example 2. Refer to Figure 1 and Figure 3 . Based on the above example, in step S2, the human acupoint image is enhanced; the quality of the human acupoint image for acupoint location research is improved, and the enhanced human acupoint image can more clearly display the acupoint-related features, providing better human acupoint image data for acupoint location, making it more conducive to the recognition and analysis of acupoint features; the gamma value γ of the gamma function and the fractional order α of the fractional order differential algorithm are optimized to find the optimal parameter combination (γ, α), and based on the optimal parameter combination (γ, α), the gamma function and the fractional order differential algorithm are used to perform gamma transformation and fractional order enhancement on the human acupoint image, optimizing the contrast, brightness, and texture details of the image, thereby improving the image quality, obtaining the enhanced human acupoint image, and constructing the training dataset and test dataset based on the enhanced human acupoint image; the parameter optimization specifically includes the following steps:

[0061] Step S21: Initial individual positions; determine the comprehensive index for measuring the quality of the enhanced human acupoint image, so as to evaluate the effect of image enhancement under different parameter combinations, and thus find the optimal parameter combination for enhancing the image. High-quality images help to more accurately identify acupoint features; establish the parameter search space for the gamma value γ of the gamma function and the fractional order α of the fractional order differential algorithm, and preset the number of initial individual positions H0, the minimum number of individual positions H min , the maximum number of search times T, and the fitness threshold f th . Randomly initialize H 0 individual positions within the parameter search space. Each individual position represents a set of parameter combinations. The information entropy, average gradient, and variance of the enhanced human acupoint image obtained based on the parameter combination are weighted and combined as the fitness value corresponding to the individual position; the formula used is as follows:

[0062] ;

[0063] ;

[0064] ;

[0065] ;

[0066] where f is the fitness value, S, R, and D are the information entropy, average gradient, and variance respectively, i is the gray level index, p i is the probability density function of the i-th gray level, M and N are the sizes of the human acupoint image in the vertical and horizontal directions respectively, m and n are the position indices in the vertical and horizontal directions respectively, is the pixel value of the human acupoint image at , is the gradient of the human acupoint image at , where ε1, ε2, and ε3 are the weight coefficients of information entropy, average gradient, and variance respectively;

[0067] Step S22: Update the individual positions; search more effectively for the optimal parameter combination in the parameter search space, while increasing the diversity of the search to avoid falling into local optima and ensuring that the image enhancement parameters most suitable for acupoint localization are found; update each individual position according to the difference between the oscillation coefficient and the individual position, and introduce random perturbations to enhance the diversity and exploratory nature of the search process; the formula used is as follows:

[0068] ;

[0069] ;

[0070] In the formula, and are the h-th individual positions at the (t + 1)-th and t-th searches respectively, Q t is the oscillation coefficient at the t-th search, h is the individual position index, t is the search number index, is an individual position randomly selected at the t-th search, r1 and r2 are two non-interfering random numbers within the range (0, 1), sin(·) and round(·) are the sine function and the rounding function respectively, and randn is a random number subject to a normal distribution;

[0071] Step S23: Search number detection; avoids waste of resources caused by unlimited search, and also prevents premature termination of the search and missing the optimal parameter combination, improving the efficiency and accuracy of the entire optimization of human acupoint image enhancement parameters; if the search number , then go to Step S24; otherwise, go to Step S25;

[0072] Step S24: Update the quantity; dynamically adjust the number of individual positions according to the actual situation of the search, which can make more effective use of computing resources, increase the probability of finding the optimal parameter combination, and thus provide more suitable image enhancement parameters for acupoint localization; includes the following steps:

[0073] Step S241: Calculate the evolution rate; quantify the degree of evolution in the search process, provide a basis for judging whether the search has stagnated, and thus determine whether it is necessary to adjust the number of individual positions during the optimization of image enhancement parameters; the formula used is as follows:

[0074] ;

[0075] In the formula, L t+1 is the evolution rate at the (t + 1)-th search, and They are the global optimal positions during the \(o\)th and \((o - 1)\)th searches respectively. The global optimal position is the individual position with the smallest fitness value. and are respectively and their fitness values. \(o\) is the search - times index. is the ceiling symbol. is the smoothing term. It is used to avoid a zero denominator.

[0076] Step S242: Evolution - rate detection; If the evolution rate \(L\) t+1 is less than or equal to \(10\) -6 , it indicates that the search process has tended to stagnate, then go to step S243; otherwise, go to step S25.

[0077] Step S243: Update the number of individual positions; Adjust the number of individual positions according to the search state to optimize the search process and increase the probability of finding the optimal parameter combination suitable for acupoint - location image enhancement; Calculate the new number of individual positions \(H\) based on the fitness value of the global optimal position t+1 , if \(H\) t+1 is less than or equal to \(H\) t , then sort all individual positions in ascending order according to the fitness value, retain the first \(H\) t+1 individual positions, discard the remaining individual positions, and let the search focus near the global optimal position to perform a more refined local search; otherwise, retain the original \(H\) t individual positions and randomly initialize individual positions again within the parameter search space to expand the search range and increase the chance of finding a better solution; The formula used is as follows:

[0078] ;

[0079] In the formula, \(H\) t , \(H\) t+1 and \(H\) 0.5T are the numbers of individual positions during the \(t\)th, \((t + 1)\)th, and \(0.5T\)th searches respectively. is the global optimal position during the \((t + 1)\)th search. is its fitness value. is the minimum fitness value since the \((t + 1)\)th search.

[0080] Step S25: Determine the optimal parameter combination; Update the fitness values of individual positions. When the fitness value of the global optimal position is less than the fitness threshold \(f\) thIf so, the parameter combination represented by the global optimal position is the optimal parameter combination (γ, α), and the parameter optimization is completed; otherwise, if the maximum search times are reached, return to step S21 to re-initialize the individual positions; otherwise, increment the search times by 1 and return to step S22 to continue the search.

[0081] By performing the above operations, for the problems existing in the existing acupoint location methods, such as poor quality of human acupoint images, unclear display of acupoint-related features, difficulty in accurately identifying and analyzing acupoint features, resulting in a large mis-identification rate of acupoint location, this solution optimizes the parameters of the gamma value and fractional order. According to the difference between the oscillation coefficient and the individual position, and introducing random perturbation, the individual position is updated. The quantity is continuously updated according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, the gamma function and fractional order differential algorithm are used to perform gamma transformation and fractional order enhancement on the human acupoint image to obtain the enhanced human acupoint image. Furthermore, a training data set and a test data set are constructed, which improves the quality of the human acupoint image, more clearly displays the acupoint-related features, provides reliable input for subsequent acupoint location, is conducive to the identification and analysis of acupoint features, reduces the mis-identification rate, and thus improves the accuracy of acupoint location.

[0082] Embodiment 3, refer to Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S3, an acupoint recognition model is constructed, and the YOLOv8 network architecture is used to complete the construction of the acupoint recognition model; the YOLOv8 network architecture includes a backbone network, a neck network, and a detection head; the backbone network is responsible for extracting feature maps of different scales from the enhanced human acupoint images in the training data set through convolution operations; the neck network is responsible for further fusing and processing the multi-scale feature maps extracted by the backbone network to enhance the expression ability of the features; the detection head is responsible for acupoint recognition according to the feature maps output by the neck network; in the construction of the acupoint recognition model, a wisdom feature convolution is designed to replace the standard convolution in the backbone network. The wisdom feature convolution consists of an activation recombination unit and a dual-type convolution optimization unit, and specifically includes the following steps:

[0083] Step S31: Activation recombination unit; in acupoint recognition, the key points and detailed information of the human body are usually relatively complex, and more precise feature extraction is required. The activation recombination unit helps to eliminate data noise while retaining important features, effectively improving the recognition ability of features of different scales and different importance, and avoiding the influence of some irrelevant or interfering features on the recognition accuracy. For acupoint recognition, it can focus on the key point positions in the human feature map and improve the accuracy of acupoint recognition; it includes the following steps:

[0084] Step S311: Group normalization; perform group normalization on the input feature map I; the formula used is as follows:

[0085] ;

[0086] Wherein, μ and σ are the mean and standard deviation of I respectively, φ and β are the scaling factor and offset factor respectively, is the feature map after group normalization processing, is the smoothing term, which is used to avoid the denominator being 0;

[0087] Step S312: Calculate weights; normalize the scaling factor of each channel to obtain the weight corresponding to each channel; the formula used is as follows:

[0088] ;

[0089] Wherein, U is the number of channels in I, ω u is the weight corresponding to the u-th channel, u and v are the channel indices respectively, φ u and φ v are the scaling factors of the u-th and v-th channels respectively, is the weight set;

[0090] Step S313: Channel gating activation; for acupoint recognition, structures such as joints, muscles or bones of the human body often contain differential feature information. Through channel gating activation, features with high information content can be intelligently retained while suppressing irrelevant or redundant features, thereby improving the stability and accuracy of acupoint recognition; map the weighted feature map to the interval (0, 1) through the Sigmoid function, and preset the threshold W th for gating processing. For channels greater than the threshold W th , set their weights to 1 to obtain the activation weight W1. For channels less than the threshold W th , set their weights to 0 to obtain the suppression weight W2; the formula used is as follows:

[0091] ;

[0092] Wherein, Gate(·) and Sigmoid(·) are the gating function and activation function respectively, W a is the weight, a is the index, W1 and W2 are the activation weight and suppression weight respectively, W1 = 1, W2 = 0;

[0093] Step S314: Split the input feature map; multiply I by W1 and W2 respectively, so as to split I into the feature map activation component and the feature map suppression component ; the formula used is as follows:

[0094] ;

[0095] ;

[0096] In the formula, is the element-wise multiplication operator;

[0097] Step S315: Split and recombine; For acupoint recognition, diverse features can be extracted from different channels and details, and the interaction between different features can be enhanced through splitting and recombination, thereby capturing the complex morphological information of acupoints; and are evenly split along the channels, and two sub-feature maps of equal size are evenly split from as and , and two sub-feature maps of equal size are evenly split from as and . The evenly split sub-feature maps are cross-combined to obtain two new feature maps I ω1 and I ω2 . Then, I ω1 and I ω2 are connected to obtain the composite feature map I ω ; The formula used is as follows:

[0098] ;

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] In the formula, is the uniform splitting operation along the channel, is the element summation operator, and ∪ is the connection operator;

[0104] Step S32: Dual-type convolution optimization unit; Combining dual-type convolution can fully extract the overall information and subtle changes, optimize the acupoint recognition effect, and effectively improve the accuracy of acupoint recognition in the learning of details when dealing with complex human acupoint images; It includes the following steps:

[0105] Step S321: Split the composite feature map; Split I ω according to and channels to obtain the core feature map I upper and the detail feature map I below , and then apply 1×1 convolution to compress I respectivelyupper and I below The number of channels is used to reduce channel redundancy, thereby obtaining the core feature map after convolutional compression and the detailed feature map ; where is a control parameter ;

[0106] Step S322: Dual-type convolution combination; Acupoint recognition is a task with very high requirements for position accuracy. Through this dual-type convolution optimization, when processing features of different scales, the error can be minimized to ensure accurate recognition and positioning of acupoints; for , group convolution and pointwise convolution are first performed separately and are added to obtain the core convolution feature map O1; for , pointwise convolution is first performed, and then and are connected to obtain the detailed convolution feature map O2; the formulas used are as follows:

[0107] ;

[0108] ;

[0109] where GWC(·) and PWC(·) are group convolution and pointwise convolution respectively

[0110] Step S323: Feature reconstruction and optimization; It can automatically adjust the importance of each feature map and help focus on important acupoint features according to actual needs, enabling intelligent adjustment according to the characteristics of the input image in different situations, avoiding over-reliance on a single feature, not only helping to improve the perception ability of complex acupoint features, but also reducing the interference of irrelevant features through a weighting mechanism, providing more robust acupoint recognition performance; Use global average pooling to process O1 and O2 respectively to obtain the core global feature B1 and the detailed global feature B2, and then generate the core weight factor and the detailed weight factor , and then use and to perform weighted combination on O1 and O2 respectively to obtain the feature fusion map O; the formulas used are as follows:

[0111] ;

[0112] ;

[0113] ;

[0114] Among them, AvgPooling(·) is the global average pooling function, O g , B g and are the convolutional feature map, the global feature, and the weight factor respectively, and g is the index.

[0115] By performing the above operations, for the existing acupoint location method, the human acupoint image features have multi-scale and complex local details, and the features at different scales often appear in different ways in the image. It is difficult for the existing methods to accurately extract details and make effective feature fusion at different scales, resulting in difficulties in accurately locating the acupoint positions. In this solution, an intelligent feature convolution is designed to replace the standard convolution in the backbone network. The activation and recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then weights them through global average pooling and a soft attention mechanism to obtain a feature fusion map, realizing the effective extraction of detail and overall information, thereby enhancing the perception ability of complex acupoint shapes, reducing redundant information and optimizing the fusion of multi-scale features, ensuring the accurate capture of features at different scales, effectively extracting subtle and key acupoint features from complex human acupoint images, and significantly improving the accuracy and stability of acupoint recognition.

[0116] Example 4, refer to Figure 1 . Based on the above example, in step S4, the acupoint location is to collect the human acupoint image to be located and perform image enhancement to obtain the enhanced human acupoint image to be located, then input the enhanced human acupoint image to be located into the acupoint recognition model for processing, identify the area where the acupoint is located, and perform annotation, and output the annotated human acupoint image to complete the acupoint location.

[0117] Example 5, refer to Figure 2 . Based on the above example, the acupoint location system based on image enhancement provided by the present invention includes a human acupoint image acquisition module, a human acupoint image enhancement module, a module for constructing an acupoint recognition model, and an acupoint location module;

[0118] The human acupoint image acquisition module collects the human acupoint image with acupoint annotation and sends the data to the human acupoint image enhancement module;

[0119] The human acupoint image enhancement module optimizes the parameters of the gamma value and fractional order. According to the difference between the oscillation coefficient and the individual position, and introducing random perturbation, it updates the individual position. It continuously updates the quantity according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, it uses the gamma function and the fractional order differential algorithm to perform gamma transformation and fractional order enhancement on the human acupoint image, obtains the enhanced human acupoint image, and sends the data to the acupoint recognition model construction module;

[0120] The acupoint recognition model construction module designs an intelligent feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, and then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then weights them through global average pooling and a soft attention mechanism to obtain a feature fusion map, and sends the data to the acupoint location module;

[0121] The acupoint location module collects the human acupoint image to be located and performs image enhancement, then inputs it into the acupoint recognition model for processing, identifies the area where the acupoint is located, and performs annotation, and outputs the annotated human acupoint image.

[0122] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0123] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention.

[0124] The above describes the present invention and its implementation manners. This description is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and design similar structural manners and embodiments without creative efforts without departing from the purpose of the present invention, they shall fall within the protection scope of the present invention.

Claims

1. An acupoint location method based on image enhancement, characterized in that: The method includes the following steps: Step S1: Acquisition of human acupoint images; acquiring human acupoint images with acupoint markings; Step S2: Enhancement of human acupoint images; optimizing the parameters of the gamma value γ of the gamma function and the fractional order α of the fractional differential algorithm. According to the difference between the oscillation coefficient and the individual position, and introducing random perturbations, update the individual positions. Continuously update the quantity according to the evolution rate to find the optimal parameter combination (γ, α). Based on the optimal parameter combination (γ, α), use the gamma function and the fractional differential algorithm to perform gamma transformation and fractional enhancement on the human acupoint images to obtain enhanced human acupoint images, and construct a training dataset and a test dataset based on the enhanced human acupoint images; Step S3: Construction of an acupoint recognition model; Step S4: Acupoint localization; acquire the human acupoint image to be localized and perform image enhancement, then input it into the acupoint recognition model for processing, identify the area where the acupoint is located, and perform markings, and output the marked human acupoint image; In step S2, it includes step S22: Update of individual positions; according to the difference between the oscillation coefficient and the individual position, update each individual position and introduce random perturbations; the formula used is as follows: ； ； wherein, and are the positions of the h-th individual at the (t + 1)-th and t-th searches respectively, Q t is the oscillation coefficient at the t-th search, h is the individual position index, t is the search number index, T is the maximum search number, is an individual position randomly selected at the t-th search, r1 and r2 are two non-interfering random numbers, sin(·) and round(·) are the sine function and the rounding function respectively, and randn is a random number subject to a normal distribution.

2. The acupoint positioning method based on image enhancement according to claim 1, wherein: In step S2, the enhancement of the human acupoint images optimizes the parameters of the gamma value γ of the gamma function and the fractional order α of the fractional differential algorithm to find the optimal parameter combination (γ, α). Based on the optimal parameter combination (γ, α), use the gamma function and the fractional differential algorithm to perform gamma transformation and fractional enhancement on the human acupoint images to obtain enhanced human acupoint images, and construct a training dataset and a test dataset based on the enhanced human acupoint images; The parameter optimization specifically includes the following steps: Step S21: Initial individual positions; establish a parameter search space for the gamma value γ of the gamma function and the fractional order α of the fractional order differential algorithm, and preset the number H0 of initial individual positions, the minimum number H min of individual positions, the maximum number of search times T, and the fitness threshold f th , randomly initialize H 0 individual positions within the parameter search space. Use each individual position to represent a set of parameter combinations, and weight and combine the information entropy, average gradient, and variance of the human acupoint enhancement image obtained based on the parameter combinations as the fitness value of the corresponding individual position; Step S22: Update of individual positions; Step S23: Search times detection; If the search times , then go to step S24; Otherwise, go to step S25; where t is the search times index; Step S24: Update of quantity; Step S25: Determination of the optimal parameter combination; update the fitness value of the individual position. When the fitness value of the global optimal position is less than the fitness threshold f th , then the parameter combination represented by the global optimal position is the optimal parameter combination (γ, α), and the parameter optimization is completed; otherwise, if the maximum search times are reached, return to step S21 to re-initialize the individual position; otherwise, increment the search times by 1 and return to step S22 to continue the search.

3. The acupoint location method based on image enhancement according to claim 2, wherein: In step S24, the update of the quantity specifically includes the following steps: Step S241: Calculation of the evolution rate; the formula used is as follows: ； Where, L t+1 is the evolution rate at the (t + 1)-th search, and are the global optimal positions at the o-th and (o - 1)-th searches respectively, and are and 's fitness values respectively, o is the search number index, is the ceiling symbol, is the smoothing term; Step S242: Evolution rate detection; if the evolution rate L t+1 is less than or equal to 10 -6 , then go to step S243; otherwise, go to step S25; Step S243: Update the number of individual positions; calculate the new number of individual positions H based on the fitness value of the global optimal position t +1 , if H t+1 is less than or equal to H t , then sort all individual positions in ascending order according to the fitness value, retain the first H t+1 individual positions, and discard the remaining individual positions; otherwise, retain the original H t individual positions, and randomly initialize individual positions again within the parameter search space; the formula used is as follows: ； where H t , H t+1 and H 0.5T are the number of individual positions at the t-th, (t + 1)-th, and 0.5T-th searches respectively, is the global optimal position at the (t + 1)-th search, is 's fitness value, is the minimum fitness value since the (t + 1)-th search.

4. The acupoint location method based on image enhancement according to claim 1, wherein: In step S3, the construction of the acupoint recognition model is completed using the YOLOv8 network architecture; the YOLOv8 network architecture includes a backbone network, a neck network, and a detection head; the backbone network is responsible for extracting feature maps of different scales from the enhanced human acupoint images in the training dataset through convolutional operations; the neck network is responsible for further fusing and processing the multi-scale feature maps extracted by the backbone network; the detection head is responsible for acupoint recognition based on the feature maps output by the neck network; in the construction of the acupoint recognition model, a smart feature convolution is designed to replace the standard convolution in the backbone network. The smart feature convolution consists of an activation recombination unit and a dual-type convolution optimization unit, and specifically includes the following steps: Step S31: Activation recombination unit; includes the following steps: Step S311: Group normalization; perform group normalization on the input feature map I; Step S312: Calculation of weights; normalize the scaling factors of each channel to obtain the weights corresponding to each channel; Step S313: Channel gating activation; map the weighted feature map to the interval (0, 1) through the Sigmoid function, and preset the threshold W th Perform gating processing. For channels greater than the threshold W th set their weights to 1 to obtain the activation weight W1. For channels less than the threshold W th set their weights to 0 to obtain the inhibitory weight W2; the formula used is as follows: ； where Gate(·) and Sigmoid(·) are the gating function and the activation function respectively, W a is the weight, a is the index, W1 and W2 are the activation weight and the inhibition weight respectively, W1 = 1, W2 = 0, is the set of weights, is the feature map after group normalization; Step S314: Split the input feature map; multiply I by W1 and W2 respectively, so as to split I into the feature map activation component and the feature map suppression component ; Step S315: Split and Recombine; For acupoint recognition, diverse features can be extracted from different channels and details, and the interaction between different features can be enhanced through splitting and recombination, thereby capturing the complex morphological information of acupoints; and are evenly split along the channels respectively, and two sub-feature maps of equal size are evenly split out from and , and two sub-feature maps of equal size are evenly split out from and . The evenly split sub-feature maps are cross-combined to obtain two new feature maps I ω1 and I ω2 . Then, I ω1 and I ω2 are connected to obtain the composite feature map I ω ; Step S32: Dual-type convolution optimization unit.

5. The acupoint location method based on image enhancement according to claim 4, wherein: In step S32, the dual-type convolution optimization unit specifically includes the following steps: Step S321: Split the composite feature map; split I ω according to and channels to obtain the core feature map I upper and the detail feature map I below , and then apply 1×1 convolution to compress the number of channels of I upper and I below respectively, so as to obtain the core feature map and the detail feature map after convolution compression; where is a control parameter, and U is the number of channels in I; Step S322: Dual-type convolution combination; For , perform group convolution and pointwise convolution respectively first, and then add and to obtain the core convolution feature map O1; For , perform pointwise convolution first, and then connect and to obtain the detail convolution feature map O2; Step S323: Feature reconstruction and optimization; using global average pooling to process O1 and O2 respectively to obtain the core global feature B1 and the detailed global feature B2, and then generating the core weight factor and the detailed weight factor , then using and to perform weighted combination on O1 and O2 respectively to obtain the feature fusion graph O.

6. The acupoint positioning method based on image enhancement according to claim 1, wherein: In step S4, the acupoint positioning is to collect the human acupoint image to be positioned and perform image enhancement to obtain the enhanced human acupoint image to be positioned, then input the enhanced human acupoint image to be positioned into the acupoint recognition model for processing, identify the area where the acupoint is located, and perform annotation, and output the annotated human acupoint image to complete the acupoint positioning.

7. An acupoint positioning system based on image enhancement, which is used to implement the acupoint positioning method based on image enhancement according to any one of claims 1-6, characterized in that: It includes a human acupoint image acquisition module, a human acupoint image enhancement module, a module for constructing an acupoint recognition model, and an acupoint positioning module; The human acupoint image acquisition module collects the human acupoint image with acupoint annotation and sends the data to the human acupoint image enhancement module; The human acupoint image enhancement module optimizes the parameters of the gamma value and fractional order, based on the difference between the oscillation coefficient and the individual position, and introduces random perturbation to update the individual position. It continuously updates the quantity according to the evolution rate to find the optimal parameter combination. Based on the optimal parameter combination, it uses the gamma function and fractional order differential algorithm to perform gamma transformation and fractional order enhancement on the human acupoint image to obtain the enhanced human acupoint image, and sends the data to the module for constructing the acupoint recognition model; The module for constructing the acupoint recognition model designs an intelligent feature convolution to replace the standard convolution in the backbone network. The activation recombination unit intelligently screens the features through channel gating activation, then splits and recombines the feature map to obtain a composite feature map. The dual-type convolution optimization unit splits the composite feature map and combines group convolution and pointwise convolution to obtain a core convolution feature map and a detail convolution feature map, and then performs weighting through global average pooling and soft attention mechanism to obtain a feature fusion map, and sends the data to the acupoint positioning module; The acupoint positioning module collects the human acupoint image to be positioned and performs image enhancement, then inputs it into the acupoint recognition model for processing, identifies the area where the acupoint is located, and performs annotation, and outputs the annotated human acupoint image.

Images