Traditional Chinese medicine acupuncture positioning method and system based on image processing

By analyzing the directional changes and posture deviations between skeleton nodes, stable structural segments are screened. Combining proportional trends and abrupt change point identification, and employing bidirectional mapping error comparison, the problem of low acupoint recognition accuracy in traditional Chinese medicine acupuncture positioning is solved, achieving accuracy and consistency in acupoint positioning.

CN121366205APending Publication Date: 2026-01-20WUGANG PHARMACEUTICAL CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202511535261.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-27
Publication Date
2026-01-20

AI Technical Summary

Technical Problem

Traditional Chinese medicine acupuncture positioning techniques ignore the dynamic factors of posture changes on skeletal structure, resulting in low acupoint recognition accuracy, high risk of structural anchor point deviation, and failure to reflect local proportional change trends during affine mapping, affecting the continuity and accuracy of acupoint positioning.

Method used

By analyzing the directional changes and posture deviations between skeletal nodes, stable structural segments are screened. Combining proportional trends and abrupt change point identification, bidirectional mapping error comparison is used to optimize the fitting of bone segment structures. By using sliding window merging and label annotation, the coherence and structural clarity of acupoint positioning are achieved.

Benefits of technology

It improves the accuracy and consistency of acupoint location, enhances the ability to distinguish movement states, improves the adaptability of graphic coordinates and template matching, and ensures the accuracy of acupoint location and the continuity of structure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121366205A_ABST
    Figure CN121366205A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of image analysis, in particular to a traditional Chinese medicine acupuncture positioning method and system based on image processing, and the method comprises the following steps: collecting a posture image, analyzing the deviation and motion state of an included angle, screening a bone segment with a stable structure, mapping acupuncture point anchoring input, recognizing the sudden change of a proportion, and adjusting the coordinates of the bone segment. And after affine errors are compared, graph mapping information is obtained, and acupoint path output annotation records are segmented. According to the method, by analyzing the posture deviation and the bone chain dynamic characteristics, the judgment capacity for the motion state is enhanced, the stable structure is screened through the bone segment included angle change trend, the acupoint anchoring precision is improved, the accuracy of bone segment structure fitting is optimized by combining the proportion trend and mutation point recognition, and the accuracy of bone segment structure fitting is improved by adopting bidirectional mapping error comparison and type classification. According to the method, the adaptability of graph coordinates and template matching is improved, continuous segmentation and structure recognition of path paragraphs are achieved through sliding window merging and label labeling, acupuncture point positioning is more coherent, and the structure is clear.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of image analysis, in particular to a traditional Chinese medicine acupuncture positioning method and system based on image processing. BACKGROUND

[0002] The technical field of image analysis includes extraction, identification and understanding of information contained in an image, including detection of image features, segmentation and identification of image targets, labeling and analysis of key points or structures in the image, etc. It is commonly applied to face recognition, medical image analysis, industrial detection, automatic driving and other scenes. In the image analysis process, image preprocessing, feature extraction, morphological modeling and data labeling are involved. The technical path includes specific operation modes based on geometric feature modeling, texture recognition or color distribution, etc. The overall technical development trend of image analysis tends to be fine recognition and structured extraction, emphasizing semantic understanding and precise feature positioning in complex scenes. It is usually closely related to computer vision, image processing algorithms and target detection methods. The traditional Chinese medicine acupuncture positioning method based on image processing refers to obtaining a human body surface image by using an image acquisition device, positioning the position of a traditional Chinese medicine acupoint in an image coordinate system through image coordinate transformation, feature point recognition and acupoint space mapping, etc. It covers standard posture acquisition of human body images, extraction of key anatomical points and derivation of acupoint reference points. Specifically, a standard acupoint template is established and matched with bone point features in the image, and two-dimensional image coordinates and acupoint model coordinates are mapped to complete automatic labeling of traditional Chinese medicine acupoints in image space. The positioning method uses image acquisition, image feature detection, acupoint mapping calculation, combines a preset human acupoint database and anatomical point index rules, and performs corresponding point calculation and acupoint number generation.

[0003] In the traditional Chinese medicine acupuncture positioning technology, only static image features are used for acupoint recognition, dynamic factors such as the influence of posture changes on the skeleton structure are ignored, there is a lack of stability recognition mechanism for continuous structure of bone chains, and the continuous direction trend cannot be extracted in the angle turning of multiple bone segments, resulting in the risk of deviation of structure anchor points. In the image coordinate and acupoint template matching process, the local scale change trend is not fully utilized for bone segment adjustment, so that the fitting ability near the mutation point is weak. In the affine mapping process, one-way error analysis is relied on, and the integrity of structure deviation cannot be reflected. In the path recognition, there is a lack of integrated judgment of the continuous change trend of acupoint sequence, causing the dispersion of boundary point detection in the sliding window, and affecting the complete labeling of subsequent paragraph index. SUMMARY

[0004] The purpose of the present application is to solve the problems existing in the prior art, and to provide a traditional Chinese medicine acupuncture positioning method and system based on image processing.

[0005] In order to achieve the above object, the present application adopts the following technical scheme: A traditional Chinese medicine acupuncture positioning method based on image processing, comprising the following steps: S1: using an image acquisition device, analyzing the direction change of the connection line between the skeleton nodes, calculating the angle offset and tracking the space turning path of the connection line, judging the motion state in combination with the length change speed of the bone chain, constructing the posture deviation type label, outputting the posture adjustment prompt information, and obtaining the posture image acquisition result; S2: using the posture image acquisition result, screening the bone segment connection lines with consistent directions, judging the bone segment set range with consistent angle change directions, screening the structure stable section, and mapping it as the anchor and matching input of acupoint positioning, obtaining the structure continuous angle interval value; S3: using the structure continuous angle interval value, obtaining the spatial distance between each skeleton node in the standard human skeleton template and comparing it with the image, analyzing the bone segment proportion trend slope, detecting the mutation segment position, identifying the angle of the upstream and downstream nodes of the mutation point, adjusting the midpoint coordinates of the bone segment, and generating the body shape stretching adjustment configuration; S4: according to the body shape stretching adjustment configuration, comparing the forward and reverse mapping errors, constructing the error point set, judging the edge length and angle change trend of the error area, classifying the error type, adjusting the affine sub-block splitting configuration, and obtaining the graph positioning mapping information; S5: according to the graph positioning mapping information, calculating the distance sequence between acupoints, constructing the continuous distance variation sequence based on the sliding window, judging the distance change trend of the window segment, merging it into a path section, extracting the boundary node, matching the paragraph index label, and outputting the structure analysis and labeling record.

[0006] As a further scheme of the present application, the posture image acquisition result includes the bone chain direction offset path, the bone chain cumulative change trajectory, and the posture deviation type label, the structure continuous angle interval value includes the stable direction section number, the anchor point set, and the angle change trajectory, the body shape stretching adjustment configuration includes the proportion mutation segment position index, the structure adjustment angle group, and the target node coordinate update table, the graph positioning mapping information includes the affine error partition number, the coordinate mapping residual record, and the sub-block splitting structure index, and the structure analysis and labeling record includes the paragraph boundary coordinate list, the path number label set, and the window segmentation index set.

[0007] As a further scheme of the present application, the acquisition step of the posture image acquisition result is specifically: S111: using an image acquisition device, acquiring user images in real time, extracting skeleton key node coordinates, generating bone chain connection lines by connecting adjacent bone points, analyzing the direction vector of the continuous connection line, calculating the direction angle difference between each pair of connection lines, judging the node sequence with consistent direction changes, and obtaining the continuous section with consistent angle changes; S112: According to the continuous section of the included angle change, the corresponding bone chain connecting line is extracted, the direction angle difference of each pair of adjacent bone chain sections is analyzed according to the node coordinates and the bone section length, the spatial turning total offset path of the corresponding connecting line in the image coordinate system is tracked, and the direction disturbance intensity coefficient is calculated; S113: The motion state is distinguished by calling the direction disturbance intensity coefficient and combining the change speed of the current bone chain length, the posture deviation type label is constructed, the posture adjustment prompt information is output, and the posture image acquisition result is obtained.

[0008] As a further scheme of the present application, the acquisition step of the structure continuous included angle interval value is specifically: S211: The posture image acquisition result is obtained, the continuous connecting section in the skeleton structure is extracted, the direction vector of each section is analyzed, the bone section set with consistent direction change is screened, and the direction continuous section data is generated; S212: According to the direction continuous section data, the direction included angle change between adjacent bone sections is detected, the included angle change amplitude is counted and interval classified, the included angle continuous change state is established, and the included angle continuous change interval is obtained; S213: The structure fluctuation characteristics of each group of included angle intervals are judged by calling the included angle continuous change interval, the structure stable section is screened, and the anchor and matching input for acupoint positioning is mapped, and the structure continuous included angle interval value is obtained.

[0009] As a further scheme of the present application, the acquisition step of the body shape stretching adjustment configuration is specifically: S311: The structure continuous included angle interval value is called, the spatial distance between each skeleton node in the standard human skeleton template is obtained, and compared with the corresponding section of the image, the proportion relationship of each section is calculated, and the bone section proportion relationship sequence is obtained; S312: According to the bone section proportion relationship sequence, the proportion change trend of continuous bone sections is analyzed, the slope sequence of proportion change is extracted, the direction reversal position in the slope trend is detected, the connection angle of the upstream and downstream nodes of the mutation point is identified, and the proportion change mutation feature value is generated; S313: The spatial coordinates of the middle position points of the bone sections are adjusted by calling the proportion change mutation feature value, the local bone section morphology is reconstructed, and the body shape stretching adjustment configuration is established.

[0010] As a further scheme of the present application, the acquisition step of the graphic positioning mapping information is specifically: S411: According to the body shape stretching adjustment configuration, the affine mapping error from the template coordinates to the image coordinates is calculated, and the reverse mapping comparison of the image coordinates to the template coordinates is performed, the affine mismatch degree of each node is calculated, and the error point set is constructed according to the mismatch distribution; S412: Call the error point set, judge the edge length extension direction and the angle change amplitude of the error continuous region by calculating the bone segment length change rate and the angle change rate between adjacent acupoint points, classify the error type, including linear elongation, torsional deformation, nonlinear diffusion, and generate error morphology feature distribution; S413: According to the error morphology feature distribution, according to the distribution characteristics of the error region, including the region arrangement density and the center offset trend, adjust the split boundary of the affine sub-block, and obtain the graphic positioning mapping information.

[0011] As a further scheme of the present application, the structure analysis and labeling record acquisition step is specifically: S511: Call the graphic positioning mapping information, calculate the distance between adjacent coordinate points on the acupoint path, construct a continuous distance variation sequence based on a sliding window, and obtain an acupoint distance sequence; S512: According to the acupoint distance sequence, judge the consistency of the distance change amplitude in each window segment, merge adjacent windows into continuous segments, and obtain continuous segment distribution data; S513: Call the continuous segment distribution data, extract the start and end acupoint coordinate points of each segment, construct the segment boundary, assign the segment index label, and output the structure analysis and labeling record.

[0012] A traditional Chinese medicine acupuncture positioning system based on image processing, which is used to execute the above-mentioned traditional Chinese medicine acupuncture positioning method based on image processing, the system comprises: The posture acquisition and analysis module uses an image acquisition device to analyze the direction change of the connection line between the skeleton nodes, calculates the angle offset and tracks the space turning path of the connection line, judges the motion state in combination with the bone chain length change speed, constructs a posture deviation type label, outputs a posture adjustment prompt information, and obtains a posture image acquisition result; The bone segment continuity screening module uses the posture image acquisition result to screen the bone segment connection lines with consistent directions, judges the bone segment set range with consistent angle change directions, screens the structure stable segments, and maps them as anchor and matching inputs for acupoint positioning, to obtain structure continuous angle interval values; The proportion mutation recognition module uses the structure continuous angle interval values to obtain the spatial distance between each skeleton node in the standard human skeleton template and compares it with the image, analyzes the proportion trend slope of the bone segment, detects the mutation segment position, recognizes the angle of the upstream and downstream nodes of the mutation point, adjusts the midpoint coordinates of the bone segment, and generates a body shape stretching adjustment configuration; The affine error determination module compares the forward and reverse mapping errors according to the body shape stretching adjustment configuration, constructs an error point set, judges the edge length and angle change trend of the error region, classifies the error type, adjusts the affine sub-block split configuration, and obtains the graphic positioning mapping information; The path section segmentation module calculates an acupoint distance sequence according to the graphic positioning mapping information, constructs a continuous distance variation sequence based on a sliding window, judges a window section distance variation trend, merges into a path section, extracts a boundary node, matches a paragraph index label, and outputs structure analysis and annotation records.

[0013] Compared with the prior art, the application has the advantages and positive effects that: In the application, the discrimination ability for the motion state is enhanced by analyzing the posture deviation and the bone chain dynamic characteristics, the precision of acupoint anchoring is improved by using the bone segment angle change trend to screen stable structures, the accuracy of bone segment structure fitting is optimized by combining the proportion trend and the mutation point recognition, the adaptability of graphic coordinates and template matching is improved by using bidirectional mapping error comparison and type classification, and the continuous segmentation and structure recognition of path paragraphs are realized by sliding window merging and label annotation, so that the acupoint positioning is more coherent and the structure is clear. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 It is a work flow diagram of the application; Figure 2 It is a posture image acquisition result acquisition flowchart of the application; Figure 3 It is a structure continuous angle interval value acquisition flowchart of the application; Figure 4 It is a body shape stretching adjustment configuration acquisition flowchart of the application; Figure 5 It is a graphic positioning mapping information acquisition flowchart of the application; Figure 6 It is a structure analysis and annotation record acquisition flowchart of the application. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0016] In the description of the application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application. In addition, in the description of the application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0017] Referring to Figure 1 The application provides a technical solution: a traditional Chinese medicine acupuncture positioning method based on image processing, comprising the following steps: S1: using an image acquisition device, analyzing the direction change of the connection line between the skeleton nodes, calculating the angle offset and tracking the space turning path of the connection line, judging the motion state in combination with the length change speed of the bone chain, constructing the posture deviation type label, outputting the posture adjustment prompt information, and obtaining the posture image acquisition result; S2: using the posture image acquisition result, screening the bone segment connection lines with consistent directions, judging the bone segment set range with consistent angle change directions, screening the structure stable section, and mapping it as the anchor and matching input of acupoint positioning, obtaining the structure continuous angle interval value; S3: using the structure continuous angle interval value, obtaining the spatial distance between each skeleton node in the standard human skeleton template and comparing it with the image, analyzing the bone segment proportion trend slope, detecting the mutation segment position, identifying the angle of the upstream and downstream nodes of the mutation point, adjusting the midpoint coordinates of the bone segment, and generating the body shape stretching adjustment configuration; S4: according to the body shape stretching adjustment configuration, comparing the forward and reverse mapping errors, constructing the error point set, judging the edge length and angle change trend of the error area, classifying the error type, adjusting the affine sub-block splitting configuration, and obtaining the graph positioning mapping information; S5: according to the graph positioning mapping information, calculating the distance sequence between acupoints, constructing the continuous distance variation sequence based on the sliding window, judging the distance change trend of the window segment, merging it into a path section, extracting the boundary node, matching the paragraph index label, and outputting the structure analysis and labeling record.

[0018] The posture image acquisition result includes the bone chain direction offset path, the bone chain cumulative change trajectory, and the posture deviation type label. The structure continuous angle interval value includes the stable direction segment number, the anchor point set, and the angle change trajectory. The body shape stretching adjustment configuration includes the proportion mutation segment position index, the structure adjustment angle group, and the target node coordinate update table. The graph positioning mapping information includes the affine error partition number, the coordinate mapping residual record, and the sub-block splitting structure index. The structure analysis and labeling record includes the paragraph boundary coordinate list, the path number label set, and the window segmentation index set.

[0019] Referring to Figure 2 The acquisition step of the posture image acquisition result is specifically: S111: using an image acquisition device, acquiring user images in real time, extracting skeleton key node coordinates, generating bone chain connection lines by connecting adjacent bone points, analyzing the direction vector of the continuous connection line, calculating the direction angle difference between each pair of connection lines, judging the node sequence with consistent direction changes, and obtaining the continuous section with consistent angle changes; Based on the user image obtained by the image acquisition device, first, the key node coordinates of the human skeleton are extracted by the human body posture estimation unit in the image, including the head, shoulder, elbow, wrist, hip, knee and ankle, etc. The specific extraction method is to identify the feature map formed by the RGB three-channel data in the image, determine the two-dimensional pixel position of each bone point by scanning the maximum response area of the key point heat map through convolution, connect the adjacent skeleton points to form a bone chain connection line sequence. For example, in a frame of image, the coordinates of left shoulder and left elbow are , and the direction vector of the connection line is calculated by the pixel coordinate difference , and then the change of the direction vector is analyzed, that is, the direction angle difference between adjacent connection lines is calculated, such as the direction of the connection line from left elbow to left wrist , and the included angle with the previous segment is calculated by the inverse tangent function to obtain the direction angle. The direction angle formed by each connection line and the x-axis is recorded in turn, and the value sequence of is calculated one by one. By judging the sequence of continuous multiple difference values with the same sign, the node sequence with continuous and consistent direction change is confirmed, for example, the angle change of left shoulder-left elbow-left wrist is , , , it is judged that the node segment belongs to the direction change consistent section, and the system numbers and marks it as a segment angle continuous interval, which is used as the input region for subsequent posture discrimination and disturbance calculation.

[0020] S112: According to the angle change continuous section, the corresponding bone chain connection line is extracted, the direction angle difference of each pair of adjacent bone chain segments is analyzed according to the node coordinates and bone segment length, and the spatial turning total offset path of the corresponding connection line in the image coordinate system is tracked. The formula is: ; Calculate the direction disturbance intensity coefficient; wherein is the direction angle value of the first segment skeleton connection line in the image coordinate system, which is obtained by calculating the included angle between the connection line between the first skeleton point and the first skeleton point and the horizontal axis of the image, is the direction angle value of the first segment skeleton connection line in the image coordinate system, which is obtained by calculating the included angle between the connection line between the first skeleton point and the first skeleton point and the horizontal axis of the image, is the normalized length value of the first segment skeleton connection line, which is obtained by calculating the Euclidean distance between the skeleton points and and then normalized by the length of the bone chain, is the normalized length value of the skeleton connection line of the first segment, which is obtained by calculating the Euclidean distance between the two adjacent skeleton points of the first segment and normalizing the total length of the skeleton chain, is the normalized length value of the skeleton connection line of the first segment, which is obtained by calculating the Euclidean distance between the two adjacent skeleton points of the first segment and normalizing the total length of the skeleton chain, is the normalized length value of the skeleton connection line of the first segment, which is obtained by calculating the Euclidean distance between the two adjacent skeleton points of the first segment and normalizing the total length of the skeleton chain, is the normalized length value of the skeleton connection line of the first segment, which is obtained by calculating the Euclidean distance between the two adjacent skeleton points of the first segment and normalizing the total length of the skeleton chain, is the sensitivity adjustment coefficient of the skeleton region of the first segment, which is obtained by assigning rules according to the structural complexity of the anatomical region where the bone segment is located, is the sensitivity adjustment coefficient of the skeleton region of the first segment, which is obtained by assigning rules according to the structural complexity of the anatomical region where the bone segment is located, is the index number of the bone chain segment, indicating the sequential number of the current processed bone segment in the bone chain, is the total number of skeleton points in the bone chain, is the direction disturbance intensity coefficient, indicating the overall disturbance degree of the change of the direction of the bone chain in the posture after the scale and sensitivity weighting; For the identified continuous segment of the angle change, the coordinates of the two adjacent skeleton points on each connection line segment are extracted and the length is calculated, and the distance is calculated by using the Euclidean formula, for example, the nodes , , and the Euclidean distance is pixels, and the total length of all connection line segments is recorded as the total length of the bone chain , and each segment length is divided by to obtain the normalized length , for example, the segment is , and in this way, each segment and in the bone chain is obtained, and the direction angle is also calculated in the same way, and the following formula is called: ; wherein the meanings of the parameters are as follows: indicates the direction angle of the first segment of the skeleton connection line with the horizontal axis, is the direction angle of the subsequent segment, and are the normalized lengths of the first segment and the second segment, respectively, is the sensitivity adjustment coefficient, which is used to represent the response degree of the human body part where the bone segment is located to the posture change, for example, the shoulder joint takes 1.2, and the wrist joint takes 0.9, indicates the sequential index of the first segment connection line in the bone chain, is the total number of segments of the bone chain, for example, indicates that there are 5 connection line segments in the bone chain, and in the formula, indicates the square of the difference in direction angle, indicates the average of the lengths of adjacent segments for scale weighting, and finally multiplied by indicates the square of the difference in direction angle, indicates the average of the lengths of adjacent segments for scale weighting, and finally multiplied by indicates the square of the difference in direction angle, indicates the average of the lengths of adjacent segments for scale weighting, and finally multiplied by The disturbance intensity term is obtained, and the final disturbance intensity coefficient is obtained by summing all the terms and then taking the square root ; As shown in Table 1, for the 4-segment bone chain connection line, the above table data is used to calculate as follows: The first disturbance term: ; The second disturbance term: ; The third disturbance term: ; The fourth disturbance term: ; The disturbance square sum is: ; The final disturbance intensity coefficient is: ; The result shows that the direction disturbance intensity of the bone chain is 6.105. Among them, the direction disturbance intensity coefficient is an index for quantifying the consistency fluctuation of the direction change of the continuous bone segments in the human skeleton, reflecting the cumulative intensity of the direction instability trend in the local area of the bone chain structure. The larger the coefficient, the more "non-continuous turning", "abnormal direction reconstruction" or "posture inertial drift" phenomenon exists in the bone chain, which is a key quantitative feature for evaluating whether the spatial structure in the posture image deviates from the physiological natural state. It not only can support the discrimination of whether the local structure is in a stable state, but also can be used to trigger the subsequent posture adjustment instruction generation mechanism, which belongs to the basic control quantity for judging the compliance of the collected posture and identifying the abnormal structure of the bone chain in the whole image processing process of traditional Chinese medicine acupuncture positioning.

[0021] S113: Call the direction disturbance intensity coefficient, combine the change speed of the current bone chain length to judge the motion state, construct the posture deviation type label and output the posture adjustment prompt information, and obtain the posture image acquisition result; Through the disturbance intensity coefficient Combine with the change speed of the bone chain length, first calculate the total length of the bone chain in the adjacent two frames of images And , for example , the total length of the bone chain at time t is 580 pixels, 612 pixels at time t, the length change is 32 pixels, and the time interval is 0.2 seconds, then the length change speed is pixels / second, the disturbance intensity 6.105 is combined with the speed to judge whether the current posture is stable, swing or offset through the preset rule, for example, if the disturbance intensity is greater than 6.105 and the length change speed is greater than 0.2 pixels / second, it is determined that the current posture is in the swing state, and the system will output a prompt information to adjust the posture to the left or right, and if the disturbance intensity is less than 6.105 and the length change speed is less than 0.2 pixels / second, it is determined that the current posture is in the stable state, and the system will output a prompt information to keep the current posture. and speed If the pixel / second is determined to be in a dynamic deviation state, a posture label is set according to a combination condition, for example, in this case, the state is set as “shoulder forward deviation”, the system outputs the label as prompt information, calculates an adjustment suggestion vector in combination with the head-shoulder-waist node connection angle deviation result, and outputs text information such as “please straighten the shoulder back 5 degrees”, and thus the posture image acquisition process of the frame image is completed, and the skeleton coordinates, direction angle, disturbance intensity, state label and other data are stored as input content for the next stage.

[0022] Please refer to Figure 3 The step of obtaining the interval value of the continuous structure included angle is specifically as follows: S211: Obtain the posture image acquisition result, extract the continuously connected segments in the skeleton structure, analyze the direction vectors of each segment, select the bone segment set with consistent direction change, and generate direction continuous segment data; After obtaining the posture image acquisition result, a sequence of continuously connected segments is generated according to the preset connection order by traversing the set of key nodes in the skeleton structure, for example, head-neck, neck-shoulder, shoulder-elbow, elbow-wrist, etc. form a complete upper limb bone chain, and the direction vector is calculated from the coordinates of the two end points of each connecting line in sequence, for example, the coordinates of node A are , the coordinates of node B are , the direction vector is , the direction angle of all paragraphs is calculated by calculating the direction angle of all paragraphs , the direction angle is recorded in sequence , the direction angle difference between each two adjacent direction vectors is calculated , if the signs of multiple are consistent, it indicates that the direction change remains continuity, and then it is marked as a bone segment set with consistent direction, for example, indicates that the three direction segments are in the same change trend, and the system merges them into a direction continuous segment, records the starting and ending skeleton point indexes, and the corresponding angle change sequence, and generates a direction continuous segment data structure containing: starting point ID, ending point ID, direction change sequence, continuous length, maximum deviation direction, etc., which is used for subsequent included angle classification.

[0023] S212: According to the direction continuous segment data, the direction included angle change between adjacent bone segments is detected, the included angle change amplitude is counted and classified, the included angle continuous change state is established, and the included angle continuous change interval is obtained; According to the direction continuous segment data, each pair of adjacent bone segments is traversed, the direction angle , is obtained, and the included angle change amplitude is calculated as , and all Statistical collection is performed, and interval classification is performed according to the magnitude, and the change range of the included angle is divided into three intervals: , , , respectively, denoted as micro-variation, medium-variation, and large-variation states, for example, the included angle change sequence is , and the interval attribution is micro-variation, medium-variation, medium-variation, large-variation, and large-variation, respectively. In a specific implementation, the number of each bone segment pair is constructed, the included angle change range is recorded, and the interval label is identified. The number of samples, the average value, and the extreme value in each interval are counted to construct the included angle continuous change state table structure. The sequence index is retained according to the position in the bone chain, and finally the included angle continuous change interval is formed. As shown in Table 2, the bone segment pairs with a larger included angle change range are mainly concentrated in the 1-2 and 3-4 segments, and the intermediate included angle changes significantly, which can be used as a significant region of bone segment activity. The segment 5-6 with a smaller included angle change is identified as a stable connection segment, which has structural continuity.

[0024] S213: Call the included angle continuous change interval, judge the structural fluctuation characteristics of each group of included angle intervals, filter the structural stable segments, and map them as the anchor and matching input of acupoint positioning to obtain the structural continuous included angle interval value; Call the included angle continuous change interval, analyze whether the change type in each group of included angle sequences has regularity, and calculate the included angle difference First-order difference calculation If the difference result tends to zero, that is, all the included angle change ranges remain stable, it is judged as a structural stability characteristic, for example , the difference is , and it is recorded as a stable segment. If there is a large fluctuation such as , the difference is , it is identified as an unstable segment. The structural stable segment is assigned a "stable segment" label, and the start and end bone point indexes and the included angle interval are recorded as the candidate input for anchor point selection in acupoint positioning. At the same time, the mapping structure information is constructed, the image coordinate segment corresponding to the stable segment connection line is compared with the acupoint template coordinates, and the region that can be used for subsequent position matching and mapping is confirmed. Finally, the structural continuous included angle interval value array is obtained in the posture image, and the format is Each item is labeled with the angle difference value and the stability label, which is used for deformation judgment in the subsequent body type fitting stage.

[0025] Please refer to Figure 4 , and the obtaining steps of the body type stretching adjustment configuration are as follows: S311: Call the structure continuous angle interval value, obtain the spatial distance between each skeleton node in the standard human skeleton template, and compare it with the corresponding segment of the image, calculate the corresponding proportion relationship of each segment, and obtain the bone segment proportion relationship sequence; After calling the structure continuous angle interval value, for each continuous bone segment pair, the theoretical spatial length of the corresponding segment in the standard human skeleton template is obtained, and the skeleton segment "shoulder-elbow" in the standard template is 50 pixels, "elbow-wrist" is 60 pixels, "hip-knee" is 70 pixels, "knee-ankle" is 65 pixels, etc. The image coordinates of these corresponding bone segments are extracted in the posture image respectively, and the actual spatial distance is calculated, such as the shoulder point in the image is , the elbow point is , and the segment length is pixels, and all image segment lengths are calculated in the same way. The image segment length is divided by the standard segment length to obtain the proportion relationship of each segment, for example , such as the segment is , the bone segment proportion relationship sequence is obtained as the basic data input for subsequent proportion trend analysis.

[0026] S312: According to the bone segment proportion relationship sequence, analyze the proportion trend of continuous bone segments, extract the slope sequence of proportion change, detect the direction reversal position in the slope trend, identify the connection angle of the nodes upstream and downstream of the mutation point, and generate the proportion change mutation feature value; Obtain the bone segment proportion relationship sequence, and calculate the proportion change slope of adjacent bone segments , the slope sequence is , wherein the reversal point is determined according to the change of slope direction from negative to positive or positive to negative, for example, the slope of the 4th segment to the 5th segment changes from to , so "4-5" is considered as a direction reversal segment, which is recorded as a proportion change mutation point. The connection angle of the bone segments on both sides of the mutation point is extracted as a geometric feature, for example, the mutation segment is "knee-ankle", the direction angle of the last segment "hip-knee" is , the direction angle of the next segment "ankle-foot" is , and the included angle change is , so the included angle of the mutation point is recorded as , and the mutation feature value combination is a three-tuple , and finally the proportion change mutation feature value sequence is generated, such as: ; Table 3 lists the comparison results of image segment and template segment length and proportion change, wherein "4-5" and "5-6" are direction reversal points, which are the positions of slope mutation.

[0027] S313: Call the proportional change mutation feature value, adjust the spatial coordinates of the middle position point of the bone segment, reconstruct the local bone segment morphology, and establish the body shape stretching adjustment configuration; The proportional change mutation feature value sequence is invoked, and local reconstruction processing is performed on the upstream and downstream connection segments of the mutation point at each mutation location. The connection angle and slope magnitude are used as the basis for offset adjustment. If the slope change magnitude is greater than 0.05 and the angle change is significant, the offset adjustment is applied. This triggers the coordinate repositioning process of intermediate nodes, resetting the midpoint coordinates using a linear adjustment method. For example, if a mutation point is the midpoint of the 4th segment, and the starting point of the current bone segment is... The destination is Midpoint coordinates Initially the mean Based on the mutation slope , angle difference Adjust the ratio; if the weight adjustment coefficient is set to 0.5, then the offset is... The corrected coordinates are Finally, the original midpoint coordinates are replaced to complete the bone segment reconstruction. The new coordinate set of all changed segments is recorded and combined to generate a body shape stretching adjustment configuration table. The structure includes fields such as bone segment number, original proportion, adjustment coordinates, offset, and mutation index, which serve as the input structure for subsequent graphic affine adjustment and mapping.

[0028] Please see Figure 5 The specific steps for obtaining graphic positioning mapping information are as follows: S411: Adjust the configuration based on body shape stretching, calculate the affine mapping error from template coordinates to image coordinates, and perform a reverse mapping comparison from image coordinates to template coordinates, using the formula: ; Calculate the degree of affine mismatch for each node, and construct a set of error points based on the mismatch distribution; in, For the first The degree of affine mismatch for each node is obtained by performing a difference operation on the forward and reverse affine vectors of that node. For the first The normalized positive mapping vector from template coordinates to image coordinates for each node is obtained by affine mapping of template coordinates to image coordinates and then normalizing it to a unit vector. For the first The normalized inverse mapping vector from image coordinates to template coordinates for each node is obtained by inverse affine mapping from image coordinates to template coordinates and then performing unit vector normalization. For the first The magnitude of the positive mapping vector of each node. For the first The module length of a node reverse mapping vector, is a small positive constant for preventing denominator zero, taking empirical value setting, is a node index number; According to the body size stretching adjustment configuration, a key node point set under the template coordinate system is selected, affine transformation is respectively performed on these nodes, forward mapping vectors from the template coordinate system to the image coordinate system are obtained, and inverse affine transformation is performed on the image coordinate system to obtain reverse mapping vectors , for each node , the module length of the forward and reverse unit vectors is calculated respectively , and the difference between the two vectors is squared and the error is calculated, and finally the affine mismatch degree is calculated by substituting the following formula: ; ; Wherein, the parameters are explained as follows: is a normalized vector after template coordinate forward mapping, is a normalized vector after image coordinate reverse mapping, the module length part is the length of each vector, is a small positive constant for preventing zero division error; represents the vector norm or numerical difference, represents the Euclidean square error; As shown in Table 4, for node 3, the affine mismatch error is the largest, and each parameter is substituted: The first error: ; The second error: ; The comprehensive error: ; ​​The affine mismatch degree is a numerical index for measuring the error accumulation in the space mapping process between the template coordinate system and the image coordinate system, and is specifically reflected in the coupling degree of the direction difference and the scale difference exhibited by a single node under forward and reverse affine mapping. This parameter, with each point as the granularity, accurately describes the twisting, stretching, and inconsistent scaling of the mapping transformation in the local area, and can significantly reveal the path distortion caused by factors such as nonlinear deformation, attitude error, and image occlusion. By introducing this parameter into the system, high-robustness error positioning and regional decoupling reconstruction between the image space and the standard template space can be achieved, providing a direct basis for high-quality output of graphical positioning mapping information. Therefore, the node is recorded as an error point and included in the error point set. The error set structure records each node number, forward and reverse vectors, length difference, normalized difference, etc., for subsequent distribution pattern judgment.

[0029] S412: Call the error point set, calculate the bone segment length change rate and the angle change rate between adjacent acupoint points, judge the edge length extension direction and the angle change amplitude of the error continuous region, classify the error type, including linear elongation, twisting deformation, and nonlinear diffusion, and generate the error pattern feature distribution; According to the error point set, a bone segment connecting line segment is constructed between the error points, and the bone segment length difference and the angle difference between adjacent error points are calculated. The image coordinates of the nodes are , and the Euclidean distance is . If the template bone segment length corresponding to the two nodes is , and the image length is , then the length change rate is . If , it is marked as "linear elongation". If the angle change is distributed alternately, it is marked as "twisting deformation". If and do not have monotonic characteristics, and the overall presents irregular fluctuations, it is recorded as "nonlinear diffusion". The system constructs the error pattern feature distribution structure, records each error feature, error type, edge length change vector, angle sequence, error center offset trend, etc. for reference for the next affine sub-block adjustment.

[0030] S413: According to the error pattern feature distribution, according to the distribution characteristics of the error region, including the region arrangement density and the center offset trend, adjust the splitting boundary of the affine sub-block, and obtain the graphical positioning mapping information. According to the error shape feature distribution data, the distribution characteristics of each error region are extracted, the arrangement density of error points in the image coordinate system is counted, the number of error points in a unit area is greater than 4 points per square pixel is defined as a high-density area, if the center of gravity of the error center point is offset more than 15% of the total length of the skeleton from the center point of the template, it is judged that there is a center offset trend, the affine sub-block boundary is automatically divided according to the error region shape, the original global affine region is divided into multiple local affine sub-blocks, each sub-block satisfies the set condition that the relative error mean square deviation is less than 0.02, the affine mapping function of each sub-block is reconstructed, and new graphic positioning mapping information is output, which includes sub-block number, mapping center point, affine matrix, node distribution index, etc., which is used for subsequent accurate acupoint positioning unit calling.

[0031] Please refer to Figure 6 The structure analysis and annotation record acquisition step is specifically: S511: Call the graphic positioning mapping information, calculate the distance between adjacent coordinate points on the acupoint path, construct a continuous distance variation sequence based on a sliding window, and obtain an acupoint distance sequence. After calling the graphic positioning mapping information, first, according to the position index of each acupoint in the image coordinate system, an acupoint path point sequence is constructed , the Euclidean distance between adjacent coordinate points in the sequence is calculated to obtain a distance sequence , wherein each distance , for example, acupoint 1 is , point 2 is , and the distance is pixels. Calculate the distance of all point pairs in this way. Then set the sliding window size , and construct a continuous variation sequence in the distance sequence using a window sliding method, such as window 1 , window 2 , etc. The sliding step is 1, and the distance mean, maximum, minimum, and standard deviation of each window are recorded to form an acupoint distance variation structure, which is prepared for subsequent segment consistency judgment.

[0032] S512: According to the acupoint distance sequence, judge the consistency of the distance variation amplitude in each window segment, merge adjacent windows into continuous segments, and obtain continuous segment distribution data; According to the acupoint distance sliding window sequence, the distance fluctuation amplitude in each window segment is analyzed, the standard deviation is calculated, and it is judged whether it meets the consistency judgment condition. The consistency judgment condition is set as: the standard deviation of the current window is pixels, and the difference between the maximum and minimum values is If the condition is met, the current window is recorded as a stable window. For multiple consecutive stable windows, perform merging processing, such as windows 1 to 3 all meeting the conditions, which are merged into a continuous segment with a length of 5, the starting point of the merged segment is the starting point acupoint number of window 1, and the ending point is the ending point acupoint number of window 3, and a structure record is generated, including the starting point index, the ending point index, the average distance, the number of windows, the standard deviation in the segment, and other information. Finally, all continuous segment distribution data is obtained, which is used to determine the image range boundary of the structure continuity region.

[0033] S513: Call the continuous segment distribution data, extract the starting point and ending point acupoint coordinate points of each segment, construct the segment boundary, assign a segment index label, and output the structure analysis and labeling record; Call the continuous segment distribution data, and extract the starting point and ending point acupoint coordinate points recorded for each segment, for example, the starting point of a segment is , and the ending point is , then the segment boundary vector is , and the length, direction angle, and number of acupoints contained are recorded, and the segment number is indexed, such as S_A1, S_A2, etc. Each segment annotation information, starting and ending point index, boundary coordinates, and number label are packaged into a structure analysis and labeling record table, and the structure body contains fields such as segment ID, starting point coordinates, ending point coordinates, segment length, segment label, and acupoint sequence index. The structure output is provided to subsequent analysis, interactive display, and acupoint recognition auxiliary system for calling.

[0034] A traditional Chinese medicine acupuncture positioning system based on image processing, the traditional Chinese medicine acupuncture positioning system based on image processing is used to execute the above-mentioned traditional Chinese medicine acupuncture positioning method based on image processing, and the system comprises: The posture acquisition and analysis module uses an image acquisition device to analyze the direction change of the connection line between the skeleton nodes, calculates the angle offset and tracks the spatial turning path of the connection line, judges the motion state in combination with the length change speed of the bone chain, constructs a posture deviation type label, outputs a posture adjustment prompt information, and obtains a posture image acquisition result; The bone segment continuity screening module uses the posture image acquisition result to screen the bone segment connection lines with consistent directions, judges the bone segment set range with consistent angle change directions, screens the structure stable segments, and maps them as anchor and matching inputs for acupoint positioning to obtain structure continuous angle interval values; The proportion mutation recognition module uses the structure continuous angle interval values to obtain the spatial distance between each skeleton node in the standard human skeleton template and compares it with the image, analyzes the proportion trend slope of the bone segment, detects the mutation segment position, recognizes the upstream and downstream node angles of the mutation point, adjusts the midpoint coordinates of the bone segment, and generates a body shape stretching adjustment configuration; The affine error judgment module compares the forward and reverse mapping errors according to the body shape stretching adjustment configuration, constructs an error point set, judges the side length and included angle change trend of the error region, classifies the error type, adjusts the affine sub-block splitting configuration, and obtains the graphic positioning mapping information; The path section cutting module calculates the distance sequence between acupoints according to the graphic positioning mapping information, constructs a continuous distance change sequence based on a sliding window, judges the distance change trend of the window section, merges into a path section, extracts a boundary node, matches a paragraph index label, and outputs a structure analysis and labeling record.

[0035] The above is only a preferred embodiment of the present application, and does not limit the present application in other forms. Any person skilled in the art can use the disclosed technical content to make changes or modifications into equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application still belongs to the protection scope of the present application technical solution.

Claims

1. An image processing-based traditional Chinese medicine acupuncture positioning method, characterized in that, The method comprises the following steps: S1: using an image acquisition device, analyzing the direction change of the connection line between the skeleton nodes, calculating the angle deviation and tracking the space turning path of the connection line, combining the length change speed of the bone chain to judge the motion state, constructing the posture deviation type label, outputting the posture adjustment prompt information, and obtaining the posture image acquisition result; S2: using the posture image acquisition result, filtering the bone segment connection lines with consistent directions, judging the bone segment set range with consistent angle change directions, filtering the structure stable section, and mapping it as the anchor and matching input of acupoint positioning to obtain the structure continuous angle interval value; S3: using the structure continuous angle interval value, obtaining the space distance between each skeleton node in the standard human skeleton template and comparing it with the image, analyzing the proportion trend slope of the bone segment, detecting the mutation segment position, identifying the angle of the upstream and downstream nodes of the mutation point, adjusting the midpoint coordinates of the bone segment, and generating the body shape stretching adjustment configuration; S4: according to the body shape stretching adjustment configuration, comparing the forward and reverse mapping errors, constructing the error point set, judging the edge length and angle change trend of the error area, classifying the error type, adjusting the affine sub-block splitting configuration, and obtaining the graphic positioning mapping information.

2. The image processing based Chinese acupuncture positioning method according to claim 1, characterized in that, The posture image acquisition result includes the bone chain direction deviation path, the bone chain cumulative change trajectory, and the posture deviation type label. The structure continuous angle interval value includes the stable direction section number, the anchor point set, and the angle change trajectory. The body shape stretching adjustment configuration includes the proportion mutation segment position index, the structure adjustment angle group, and the target node coordinate update table. The graphic positioning mapping information includes the affine error partition number, the coordinate mapping residual error record, and the sub-block splitting structure index.

3. The image processing based Chinese acupuncture positioning method according to claim 1, characterized in that, The acquisition step of the posture image acquisition result is specifically: S111: using an image acquisition device, real-time acquiring user images, extracting skeleton key node coordinates, generating bone chain connection lines by connecting adjacent bone points, analyzing the direction vector of the continuous connection line, calculating the direction angle difference between each pair of connection lines, judging the node sequence with consistent direction change, and obtaining the continuous section with consistent angle change; S112: according to the continuous section with consistent angle change, extracting the corresponding bone chain connection line, analyzing the direction angle difference of each pair of adjacent bone chain sections according to the node coordinates and the bone segment length, tracking the space turning total deviation path of the corresponding connection line in the image coordinate system, and calculating the direction disturbance intensity coefficient; S113: calling the direction disturbance intensity coefficient, combining the change speed of the current bone chain length to judge the motion state, constructing the posture deviation type label and outputting the posture adjustment prompt information, and obtaining the posture image acquisition result.

4. The image processing based Chinese acupuncture positioning method according to claim 3, characterized in that, The acquisition step of the structure continuous angle interval value is specifically: S211: obtaining the posture image acquisition result, extracting the continuous connection section in the skeleton structure, analyzing the direction vector of each section, filtering the bone segment set with consistent direction change, and generating the direction continuous section data; S212: according to the direction continuous section data, detecting the direction angle change between adjacent bone segments, statistically classifying the angle change amplitude, establishing the continuous angle change state, and obtaining the continuous angle change interval; S213: Call the included angle continuous change interval, judge the structural fluctuation characteristics of each group of included angle intervals, screen the structural stable section, and map it as the anchor and matching input of acupoint positioning, and obtain the structural continuous included angle interval value.

5. The image processing based Chinese acupuncture positioning method according to claim 4, characterized in that, The obtaining step of the body shape stretching adjustment configuration is specifically: S311: Call the structural continuous included angle interval value, obtain the spatial distance between each skeleton node in the standard human skeleton template, and compare it with the corresponding paragraph of the image, calculate the corresponding proportion relationship of each paragraph, and obtain the bone segment proportion relationship sequence; S312: According to the bone segment proportion relationship sequence, analyze the proportion change trend of the continuous bone segment, extract the slope sequence of the proportion change, detect the direction reversal position in the slope trend, identify the connection angle of the upstream and downstream nodes of the mutation point, and generate the proportion change mutation feature value; S313: Call the proportion change mutation feature value, adjust the spatial coordinates of the middle position points of the bone segments, reconstruct the local bone segment morphology, and establish the body shape stretching adjustment configuration.

6. The image processing based Chinese acupuncture positioning method according to claim 5, characterized in that, The obtaining step of the image positioning mapping information is specifically: S411: According to the body shape stretching adjustment configuration, calculate the affine mapping error from the template coordinates to the image coordinates, and perform reverse mapping comparison from the image coordinates to the template coordinates, calculate the affine mismatch degree of each node, and construct an error point set according to the mismatch distribution; S412: Call the error point set, calculate the bone segment length change rate and the included angle change rate between adjacent acupoint points, judge the edge length extension direction and the included angle change amplitude of the error continuous region, classify the error types, including linear elongation, torsional deformation, and nonlinear diffusion, and generate the error morphology feature distribution; S413: According to the error morphology feature distribution, adjust the split boundary of the affine sub-block according to the distribution characteristics of the error region, including the region arrangement density and the center offset trend, and obtain the image positioning mapping information.

7. The image processing based Chinese acupuncture positioning method according to claim 1, characterized in that, The method further comprises: S5: According to the image positioning mapping information, calculate the acupoint distance sequence, construct the continuous distance variation sequence based on the sliding window, judge the distance change trend of the window segment, merge it into a path section, extract the boundary node, match the paragraph index label, and output the structure analysis and annotation record; The structure analysis and annotation record comprises a paragraph boundary coordinate list, a path number label set, and a window segmentation index set.

8. The image processing based Chinese acupuncture positioning method according to claim 7, characterized in that, The obtaining step of the structure analysis and annotation record is specifically: S511: Call the image positioning mapping information, calculate the distance between adjacent coordinate points on the acupoint path, construct the continuous distance variation sequence based on the sliding window, and obtain the acupoint distance sequence; S512: According to the acupoint distance sequence, judge the consistency of the distance change amplitude in each window segment, merge adjacent windows into continuous sections, and obtain the continuous section distribution data; S513: Call the continuous section distribution data, extract the start and end acupoint coordinate points of each section, construct the section boundary, assign the section index label, and output the structure analysis and annotation record.

9. An image processing based Chinese acupuncture positioning system, characterized in that, The system is used to implement the image processing-based traditional Chinese acupuncture positioning method according to any one of claims 1-8, and the system comprises: The posture acquisition and analysis module uses an image acquisition device to analyze the direction change of the connection line between the skeleton nodes, calculates the angle deviation, tracks the spatial turning path of the connection line, judges the motion state in combination with the length change speed of the bone chain, constructs a posture deviation type label, outputs a posture adjustment prompt information, and obtains a posture image acquisition result; The bone segment continuity screening module uses the posture image acquisition result to screen the bone segment connection lines with consistent directions, judges the bone segment set range with consistent angle change directions, screens the structure stable section, and maps it as an anchor and matching input for acupoint positioning to obtain a structure continuous angle interval value; The proportional mutation recognition module uses the structure continuous angle interval value to obtain the spatial distance between each skeleton node in a standard human skeleton template and compares it with the image, analyzes the slope of the bone segment proportion trend, detects the mutation segment position, recognizes the angle of the upstream and downstream nodes of the mutation point, adjusts the midpoint coordinates of the bone segment, generates a body shape stretching adjustment configuration, and outputs the body shape stretching adjustment configuration; The affine error determination module compares the forward and reverse mapping errors according to the body shape stretching adjustment configuration, constructs an error point set, judges the side length and angle change trend of the error area, classifies the error type, adjusts the affine sub-block splitting configuration, and obtains the graphic positioning mapping information; The path section segmentation module calculates the distance sequence between acupoints according to the graphic positioning mapping information, constructs a continuous distance variation sequence based on a sliding window, judges the distance change trend of the window section, merges it into a path section, extracts the boundary nodes, matches the paragraph index label, and outputs the structure analysis and labeling record.

Citation Information

Cited By

  • Acupuncture point positioning method and system based on artificial intelligence

    CN121724982A