183 results about "Human skeleton" patented technology

The invention discloses a physical fitness evaluation method based on human skeleton trajectory tracking, which comprises the following steps: S1, acquiring whole-process video data of a physical fitness test through video acquisition equipment, and generating a video frame according to an acquisition frequency; s2, processing the video frame by using a skeleton dynamic analysis algorithm, and extracting human skeleton key points; s3, extracting identity features of the testee through a motion map identity recognition network, and tracking and confirming the identity features; s4, performing action recognition, fragment segmentation and compliance judgment on the skeleton key point time sequence track; s5, aiming at the abnormal skeleton key points, performing complementation and time sequence smoothing by adopting an inverse kinematics inference method; s6, inputting the complete skeleton key point time sequence data into the human body action evaluation model for processing, and quantizing and outputting a physical ability evaluation index; and S7, generating a physical ability evaluation report and giving action feedback and optimization suggestions. According to the invention, objectivity, accuracy and intelligent level of physical ability evaluation are effectively improved.

The invention provides a radar-based human skeleton estimation method, system and product, and the method comprises the steps: carrying out the target detection of collected radar echo data, and generating a target four-dimensional point cloud containing distance information, speed information and angle information based on a target detection result; dividing the target four-dimensional point cloud into a plurality of human body topology point cloud blocks based on the velocity direction similarity of each point in the target four-dimensional point cloud; and inputting the human body topology point cloud block into a human body skeleton estimation network for human body skeleton estimation to obtain a human body skeleton estimation result. According to the human skeleton estimation method provided by the invention, human topology priori can be constructed from sparse millimeter wave radar point clouds, point cloud block features sensed by a structure are extracted, the modeling capability of a model for the spatial relationship of key parts of a human body is enhanced, diversified and natural daily human behaviors in a non-inductive monitoring scene can be adapted, and the human skeleton estimation accuracy is improved. The human body skeleton estimation with higher generalization ability is realized, and the accuracy and robustness of human body posture prediction can be effectively improved.

The invention discloses a dance movement auxiliary generation method and system based on three-dimensional modeling. The method comprises the following steps: constructing a three-dimensional human skeleton model; generating a joint action set of each joint node in the dance action; constructing a dance movement auxiliary model based on fusion of a time sequence convolutional network and a graph convolutional neural network, obtaining three-dimensional joint movement trajectory data in continuous frames, and obtaining relevance among dance movements by analyzing time sequence features and collaborative movement rules in the dance movements; acquiring joint actions associated with the current dance action in the joint action set, and generating a candidate action set; and calculating to obtain a connection fluency score and a posture aesthetics score with the current dance movement, and weighting to obtain the candidate movement with the highest total score as the next dance movement. The method has the advantages that through precise skeleton modeling and kinematics constraint, the deep learning technology is combined, smooth and attractive dance actions are efficiently generated, and the method is suitable for dance creation and virtual performance.

The invention relates to the technical field of body posture recognition, and particularly discloses an automatic body posture recognition system based on video analysis, which comprises a video acquisition module, a preprocessing module, a key point detection module, a skeleton tracking module, an action recognition module and a cross-view fusion module. According to the scheme, a parallel detection and information fusion mode is adopted, and the detection precision of the key points of the human body is remarkably improved by utilizing the accurate position of the high-resolution feature and the rich semantics of the low-resolution feature; the target is tracked according to the topological structure and the key point position of the human skeleton, the method has extremely high robustness for illumination and dressing changes, and a more stable and continuous human skeleton movement track is generated; the time sequence convolution and the space-time diagram convolution are combined to carry out joint modeling on the space and time double-domain features, complete action representation is obtained, and the stability, the accuracy and the adaptability of body posture recognition are effectively improved.

The invention discloses a human body behavior recognition method, system and device based on a space-time shift attention mechanism and a medium. The recognition method comprises the following steps: firstly, performing position coding on human skeleton data as input of the spatial feature extraction module; secondly, a spatial feature extraction module extracts spatial features through global cyclic shift operation, a physical topology matrix and a spatial self-attention mechanism; then, taking the output of the spatial feature extraction module as the input of a time feature extraction module, and extracting time features by the time feature extraction module through a multi-scale time convolution module and a time transformer module; then, the classification module performs classification according to the features extracted by the space and time module to obtain the output of the single-joint flow; similarly, obtaining the output of the single skeleton flow; and finally, carrying out weighted fusion on the single-joint flow output and the single-bone flow output to obtain final recognition precision. According to the method, richer local and global space-time feature information can be captured, the human skeleton topological relation is effectively supplemented, and the method has the advantage of effectively extracting the local and global space-time dependency relation between the nodes.

The application discloses a kind of spatio-temporal decoupling human behavior recognition method, device and equipment based on dynamic semantic guide mask, comprising: extracting human skeleton sequence data from original video and carrying out data enhancement;Based on dynamic semantic guide mask mechanism, the skeleton sequence after data enhancement is masked in operation in two dimensions of space and time;Skeleton sequence data after masking operation is sent into query encoder and momentum encoder respectively, spatial representation and time representation are obtained respectively, cross-domain contrast loss is constructed, contrast learning training is carried out to obtain human behavior recognition model;The video to be identified is input into human behavior recognition model, and the prediction result of human behavior in the video to be identified is obtained.The spatio-temporal decoupling human behavior recognition method, device and equipment based on dynamic semantic guide mask of the application significantly improve human behavior recognition precision and environmental adaptability without manual label.

This invention belongs to the field of computer vision technology, specifically to a method for extracting a human skeleton from point cloud data. To make the human skeleton extracted in complex environments more stable and robust, this method uses a trained deep learning network to extract features from the original point cloud. This method then aggregates these features using a feature pooling operation, deriving joint points from the aggregated features. Finally, these joint points are connected in a specific order to form a three-dimensional human skeleton, enabling applications such as behavior recognition and other human-computer interaction applications.

The invention discloses an attitude estimation method based on graph convolution and double-branch fusion, and the method comprises the steps: employing a double-branch structure, a Mama branch and a Transform branch to work in parallel, enabling the Mama branch to process long-distance time dependence information through employing an efficient state space model, capturing a long-time-span correlation mode, and enabling the Transform branch to carry out the parallel processing of the long-distance time dependence information; the Transform branch strengthens modeling of local and global attention through a self-attention mechanism and pays attention to interaction relations of different time steps, and after the two branches are output and fused, dynamic changes of human body joint points on a time sequence can be more accurately expressed, a complex time sequence mode can be flexibly and effectively processed, different human body actions can be better adapted, and posture estimation accuracy is improved. The GCN is used for extracting the spatial topological features of the human skeleton, generating the preliminary feature representation, subsequently performing spatial structure optimization on the fused time features by using the GCN, and generating the structured associated features, so that the mode of combining the spatial information and the time sequence information can more comprehensively understand the posture of the human body and consider the dynamic change of the time and the relative position and the connection relationship of the space.

The invention provides an intelligent motion generation method, an intelligent motion generation system, electronic equipment and a computer readable storage medium, in the intelligent motion generation method, motion feature data are mapped and aligned to a virtual human skeleton structure, a preliminary motion sequence is obtained, and in addition, the motion feature data are mapped and aligned to the virtual human skeleton structure; and according to the virtual human skeleton structure and the preliminary action sequence, optimizing the action of the virtual human skeleton structure, and outputting an optimized action sequence, so that after the preliminary action sequence is obtained according to a redirection strategy based on the frame sequence of the video of the human body motion, the motion of the virtual human skeleton structure is optimized. And an optimized action sequence obtained by optimizing the preliminary action sequence can be more matched with the body shape of the virtual human.

The invention discloses a swimming skeleton point coordinate data denoising method and device based on spatio-temporal topological structure learning, and belongs to the field of data processing. The method comprises the following steps: acquiring and splicing a swimming video of a target object; detecting a target object in the video, and obtaining skeleton point coordinates and corresponding confidence scores of the target object by adopting a posture estimation model; preprocessing the coordinates to obtain feature vectors; adjacency information of each skeleton point is obtained based on the human body skeleton topology, the feature vectors of all the skeleton points are input into a spatial domain noise removal network, the adjacency information of the skeleton points is aggregated, multi-scale pooling is carried out, and denoised spatial domain features are obtained; sampling the high-dimensional spatial-temporal characteristics by adopting a deformable time convolutional network; and carrying out deformable convolution operation and decoding operation on the sampling features to obtain skeleton point coordinates after time-space domain denoising. Linear and nonlinear noise in data is processed in a time-space domain in a cooperative manner, so that the limitation of a traditional low-pass filter method and an existing ST-GCN method is effectively overcome.

The embodiment of the invention provides an intelligent alarm method and device for tumble monitoring, and the method achieves the effective protection of the privacy of a user through the innovative design of a low-position visual angle collection mechanism, the setting of the installation height of a camera not exceeding the knee joint, and the starting of collection only during the abnormal motion. A detection mechanism based on a human skeleton model is constructed, and an accurate fall judgment strategy is established by combining multi-part vertical distance analysis and motion state parameters. An intelligent alarm mechanism is introduced, and quick response of an emergency contact person is realized through key video packaging and emergency channel establishment. According to the method, while the privacy of the user is protected, the defects of the traditional technology in the aspects of data acquisition, tumble detection, alarm response and the like are effectively overcome, and the practicability and reliability of tumble monitoring are remarkably improved.

The invention belongs to the field of computer vision and artificial intelligence, and particularly discloses a human body three-dimensional posture estimation method and system based on Transform and a graph convolutional network, and the method comprises the steps: firstly receiving a two-dimensional skeleton sequence, sampling a center sub-sequence, and converting the two-dimensional skeleton sequence and the center sub-sequence into an initial spatial-temporal feature containing a spatial position code; through a double-flow parallel space coding module, global dependency features are obtained through a Transform encoder, and local connection features are obtained through a graph convolutional network encoder (based on a human skeleton topological adjacent matrix); discrete cosine transformation is carried out on the complete sequence to obtain a low-frequency coefficient, and frequency domain features are generated; and after the three types of features are spliced, deep fusion is carried out through a time-frequency fusion Transform encoder, and finally, the three-dimensional attitude of the target frame is output through a regression head. The method gives consideration to both efficiency and precision, is high in robustness, and is small in calculation cost increase.

The invention discloses a two-stage limb violent behavior recognition method based on improved OpenPose and space-time diagram convolution ST-GCN, which is applied to the technical field of limb violent behavior recognition and comprises the following steps: extracting human skeleton key point time sequence data based on an improved OpenPose algorithm; wherein the improvement comprises the steps of extracting human skeleton key points based on MobileNet-V3 and carrying out global optimal matching on the human skeleton key points based on a Hungary algorithm; based on a space-time diagram convolution ST-GCN structure, analyzing and extracting space-time features of the time sequence data of the key points of the human skeleton, and outputting a limb violent behavior recognition result; wherein the space-time diagram convolution ST-GCN structure is formed by stacking a plurality of ST-GCN modules, and each ST-GCN module is composed of an attention layer, a space diagram convolution and a time diagram convolution. According to the method, the limb violent behavior identification precision is effectively improved, and the real-time performance is ensured.

The invention provides an intelligent motion generation method and system of a human body, and electronic equipment, and the method comprises the steps: obtaining motion capture data based on video data, obtaining target motion information and an initial human body skeleton key point sequence based on the motion capture data, obtaining source skeleton motion data based on a human body structure prior key point sequence of the human body skeleton, and obtaining the target motion information and the initial human body skeleton key point sequence; obtaining a virtual human image based on the target motion information, obtaining a virtual human skeleton structure based on the virtual human image, obtaining redirected skeleton data, obtaining an optical flow graph and a mask graph based on the virtual human image and the redirected skeleton data, and obtaining an initial virtual human motion image sequence based on the optical flow graph and the mask graph; and acquiring a virtual human motion image sequence based on the initial virtual human motion image sequence, and acquiring motion output of the living agent based on the virtual human motion image sequence. Therefore, coherent and natural action generation is realized.

The invention provides a practical operation process stability assessment method and device based on visual analysis, and particularly relates to the field of stability assessment, and the method comprises the steps: extracting each frame of image from a teacher practical operation video, generating a multi-channel heat map for a preset key point, constructing a two-dimensional human skeleton posture of each frame based on preset limb connection, and constructing a multi-channel heat map; and performing three-dimensional reconstruction on the continuous two-dimensional attitude sequence to obtain a three-dimensional space coordinate sequence of the key points, and calculating a comprehensive stability evaluation score based on the three-dimensional space coordinate sequence to obtain the comprehensive stability evaluation score. According to the method, the stability of limb connection in a complex shielding environment is enhanced, accurate construction of a two-dimensional posture is ensured, and an objective and accurate comprehensive evaluation result of the stability in the practical operation process can be provided.

The invention relates to the technical field of image recognition, in particular to an eight-segment brocade rehabilitation training action recognition method, which comprises the following steps of: based on an input eight-segment brocade training action image sequence, extracting human skeleton key point coordinates in each frame of image, and establishing multi-dimensional limb movement track data; according to the invention, through frame-by-frame extraction of human skeleton key points in an image sequence, multi-dimensional limb movement track data with space and time characteristics is established, and a basic structure from single-frame static analysis to cross-frame dynamic modeling is formed. By means of mirror image mapping of a left key point on the basis of a body midline, item-by-item vector operation is carried out on a mapping point and a right actual coordinate, an instantaneous deviation vector field used for measuring symmetry is constructed, and real-time quantitative evaluation of movement consistency of the left limb and the right limb is achieved. And in combination with periodic integral operation, an accumulated symmetry deviation value covering a complete action process is generated, and full-period feedback on the attitude stability and the action specification degree is provided.

The application discloses a human abnormal behavior detection method combining appearance texture and motion skeleton, and belongs to the field of computer vision; first, an original video is divided at equal intervals, and human appearance key regions in each frame and skeleton key points of each human body are extracted respectively; global motion trajectories of all human bodies are calculated; then, based on STGAT, trajectories of skeleton key points of each human body in a future frame are predicted; for each human body, predicted skeleton key points are converted into dense flow heat maps as guidance information by using DFE, and the guidance information is input into CGAN to generate human appearance key regions corresponding to skeleton key point postures; in addition, pixel-by-pixel analysis is used to eliminate background deviation of the appearance key regions to improve accuracy; finally, skeleton key point prediction values and appearance key region generation values of each human body are calculated with corresponding label values to obtain error, and the error is weighted and summed to obtain an abnormal score; the application effectively reduces the false alarm rate and realizes rapid and effective video anomaly detection.

A human whole body action real-time reconstruction method and system based on three-point sensing signals, according to the collected head and double-hand three-point sensor signals, a pre-estimated waist joint and root joint are generated through a waist joint and root joint pre-estimation algorithm; the pre-estimated waist joint and root joint are combined with a human skeleton model, and an upper body reconstruction result is generated through a reverse motion algorithm; in addition, a lower body reconstruction network based on BiGRU is used to generate a lower body reconstruction result for an input action feature sequence, and a double-threshold-based action post-processing technology is used to further stabilize the lower body action, and finally the whole body action is obtained by merging and reconstructing. The application uses only the sensing signals of three body parts of the head and the hands to pre-estimate the waist joint and the root joint, increases the number of known joints, more effectively reconstructs the upper body action and the lower body action at the same time, and significantly improves the reconstruction effect of the lower body action.

The invention provides a human skeleton behavior recognition method based on adaptive graph convolution and self-attention, and relates to the technical field of behavior recognition, and the method comprises the following steps: processing an NTU RGB + D sequence containing K types of human skeleton behaviors; constructing a spatial relationship modeling module; constructing a multi-scale time convolution module; and constructing a hybrid network. According to the technical scheme, the problems that in the prior art, an identification model is poor in robustness to complex actions and non-standard attitudes, and the identification precision is low due to insufficient spatial-temporal feature extraction are solved.

The invention discloses a human body abnormal posture detection method based on space-topology feature fusion of a skeleton, and relates to the field of computer vision. The model mainly comprises five parts, namely similarity calculation of human body skeletons, optimal matching of the human body skeletons, spatial feature extraction of the human body skeletons, topological feature extraction of the human body skeletons and feature fusion. Comprising the following steps: firstly, acquiring human body skeleton data of an input target image by using a YOLOv5 algorithm and a Lightweight-Openpose algorithm; next, performing an optimal matching process on the human body skeleton of the target image and a set template skeleton set in a corresponding scene, namely, finding a skeleton with the highest similarity with the target skeleton by calculating the similarity between the target skeleton and each skeleton in the corresponding template skeleton set; then, representing the Euclidean distance of each pair of mutually matched skeleton key points of the two skeletons in the matching process as a skeleton space vector, representing the connection relation of all skeleton key points in the target skeleton as a skeleton topological matrix, and obtaining a skeleton topological vector through GCN processing; and finally, fusing the skeleton space vector and the skeleton topology vector into a skeleton feature vector, classifying the skeleton feature vectors of all input target images by using an SVM, and carrying out human body abnormal posture detection. The process fully combines the spatial features and topological features of the human skeleton, and the accuracy of human body abnormal posture detection is improved.

The application discloses a human body skeleton generation method for behavior recognition, and steps include: 1) collecting a data set of related human body skeleton actions; 2) pre-processing the data set, selecting action category samples that need to be generated and complete action samples that need to be referenced, and dividing a training set and a test set; 3) building a skeleton style migration characteristic generation network, the generation network is composed of an encoder, a decoder and a discriminator, wherein the encoder and the decoder together constitute a generator, and the encoder is further subdivided into a content encoder and a style encoder; 4) using the training set to sequentially train the content encoder, the style encoder, the decoder and the discriminator; 5) the generated new sample is used to participate in the training of a behavior recognition network, the generalization ability of the behavior recognition network is improved, and the accuracy of the test set on the action is improved. The method has better accuracy of human body skeleton generation results and better recognition effect.

The invention discloses a human body size data determination method and device based on 3D modeling, a medium and a product. The method comprises the following steps: collecting a multi-view visible light image and a multi-view depth map of a target human body to generate multi-view point cloud data; performing alignment fusion on the multi-view point cloud data by iteratively minimizing corresponding point distance deviation among the multi-view point cloud data to generate a global point cloud model; classifying the global point cloud model to obtain a human body point cloud; performing grid reconstruction based on the human body point cloud to generate an initial three-dimensional grid model; smoothing the initial three-dimensional mesh model to obtain a final three-dimensional mesh model; and positioning a human skeleton datum point of the final three-dimensional grid model, and extracting human body size data. By implementing the technical scheme provided by the invention, the technical problems that the accuracy of the extracted key size data such as the girth and the length of the human body is insufficient and the requirement of a high-precision application scene is difficult to meet are effectively solved.

This application provides a method and apparatus for open-set skeleton action recognition based on outlier prototype learning. The method includes: constructing a neural network model; preprocessing human skeleton data by obtaining initial features through a multi-branch feature extraction network; processing the initial features using a classifier and a hypersphere feature mapper to obtain logical prediction values ​​for action categories and branch features; training the neural network model based on a training set; after the first iteration, optimizing the feature space of in-distribution samples using multi-class loss; synthesizing virtual outliers in the optimized feature space of in-distribution samples; optimizing the energy boundary by combining in-distribution samples and virtual outliers; after the second iteration, selecting the optimal model weights based on the combined performance of open-set recognition and closed-set classification on the validation set; and determining the action type and triggering the corresponding robot operation based on the energy score output by the trained neural network model and the action classification result.

The present disclosure relates to a method and device for human behavior recognition. The human behavior recognition method includes: determining a joint difference topological representation and a joint dependency topological representation based on a human skeleton sequence, and determining a graph topological feature based on the joint difference topological representation and the joint dependency topological representation, wherein the human skeleton sequence is represented as a graph structure, and the graph topological feature is the sum of the general structure topological representation, the joint difference topological representation and the joint dependency topological representation; determining the action features of the human skeleton sequence based on the human skeleton sequence and the graph topological feature; reconstructing the graph topological feature using a fine-grained prototype to obtain a refined action topological representation; determining the action category of the human skeleton sequence based on the action features of the human skeleton sequence and the refined action topological representation using a prototype contrast loss function and a cross entropy loss function. By adopting the present disclosure, it is possible to better capture subtle differences in actions and effectively extract discriminative behavior features.

The invention discloses a gait recognition method based on a Pluecker straight line, and belongs to the technical field of computer vision and biological feature recognition. The core of the method is to provide a double-layer gait feature modeling framework. The method comprises the following steps: firstly, modeling human skeletons into spatial straight lines by utilizing Pluecker coordinates, and calculating absolute rigid body motion characteristics of each skeleton between adjacent frames through dual quaternions; furthermore, by combining the absolute motion characteristics of adjacent bones, the relative motion characteristics of the joints, which can better reflect the physiological characteristics of individuals, are calculated. And finally, aggregating the two complementary features to construct a gait feature sequence, and completing identity recognition by using a time sequence deep learning model. According to the method, the overall motion information and the local joint motion information are separated and fused, so that the defects that an existing method is incomplete in feature representation and is greatly interfered by the overall motion are effectively overcome, and the accuracy, the discrimination capability and the robustness of gait recognition are remarkably improved.

The invention discloses a human body model, and relates to the technical field of teaching instruments, and the human body model comprises a first model frame which is provided with a plurality of first meshes and is used for simulating a human body skeleton; the second model frame is provided with a cavity, a first connecting part, a second connecting part and a plurality of second meshes, the first model frame is arranged in the cavity and detachably connected with the second model frame, the first connecting part is detachably connected with the second connecting part, the second model frame is used for simulating the outer contour of a human body, and the aperture of at least part of the second meshes is larger than that of the first meshes; each of the first model frame and the second model frame comprises at least one of a head and neck, a thorax, an abdomen, a pelvis and four limbs and is correspondingly arranged; the simulation structure comprises a first simulation piece and a second simulation piece, and the first simulation piece is used for simulating blood vessels, nerves and lymphatic vessels; the second simulation piece is used for simulating human muscles and organs. The problem that in the prior art, a human body model used for teaching is difficult to operate manually while the internal structure is visualized is solved.

The invention relates to the technical field of rehabilitation training, in particular to a rehabilitation training method based on multi-modal interaction and skeleton detection. The method comprises the following steps: extracting 33 key points when a human skeleton is detected through Pose model detection and a SkelegonProcessor module, calculating relative coordinates of all the key points of the human body by taking average coordinates of a left hip and a right hip as an original point of a middle hip, calculating a distance from a zero key point of an approximate nose on the top of the head to the middle hip as a trunk length, performing scaling normalization on the relative coordinates, and obtaining a trunk length; generating a human skeleton visual image; enabling the specific action to correspond to a keyboard instruction or a mouse instruction; the action and the mouse instruction or the keyboard instruction are bound and replaced through a database; two actions are set to be combined into one instruction. According to the technical scheme, whether the rehabilitation actions of the patient are standard or not and whether the joint movement range meets the requirement or not can be accurately judged in real time, instructions are sent out through action recognition, the interestingness of rehabilitation training is improved, and the rehabilitation training effect is improved.

The invention relates to a lightweight personnel wearing equipment intelligent identification method under a small sample condition. The method comprises the following steps: collecting a video frame image of an industrial production environment; carrying out human body posture estimation on the video frame image, judging whether a person exists or not, if so, extracting human body key points, and carrying out smooth processing to form human body skeleton information, and if not, re-collecting the video frame image; performing target detection on the wearing equipment area of the human body, performing fusion with set key wearing area cutting, and associating to a corresponding human body target to obtain a wearing equipment area image; carrying out wearing feature extraction on the wearing equipment area image, and constructing a wearing feature library; and obtaining the wearing features of the to-be-identified wearing equipment area image, carrying out retrieval matching on the wearing feature library, and carrying out multi-frame integration optimization by adopting a hysteresis comparison strategy to obtain a wearing identification node of the wearing equipment. Compared with the prior art, the method has the advantages of real-time light weight, accurate result and the like.

The application discloses a method, device and diagnosis and treatment machine auxiliary system for autism child behavior recognition and motion prediction, the method comprises the following steps: (1) acquiring original skeleton video data of the child from different angles; (2) inputting the acquired original skeleton video data into a view adaptive conversion unit to adaptively convert to a consistent coordinate system to obtain the optimal representation of the human skeleton; (3) inputting the optimal representation of the human skeleton into a multi-scale feature extraction unit to extract high-performance spatiotemporal features of the human skeleton; (4) inputting the high-performance spatiotemporal features into a multi-task learning unit to perform behavior recognition and motion prediction on the human body, and obtaining behavior classification results and motion prediction results. The application can predict and intervene the future motion while recognizing the behavior.

The application discloses a three-dimensional human skeleton fitting method and system based on a multi-camera joint association graph and combined search, and relates to the three-dimensional human skeleton fitting method, which comprises the following steps: S1, acquiring images collected by multiple cameras, detecting human two-dimensional joint nodes by using a two-dimensional pose estimation algorithm, and projecting the two-dimensional joint nodes into three-dimensional rays; S2, taking the joint nodes as nodes and the geometric consistency between three-dimensional rays of different cameras as edge weights, constructing a multi-camera joint node association graph; S3, performing group clustering on the cross-camera joint nodes with high geometric consistency on the association graph by using a breadth-first search, and forming a candidate skeleton cluster; S4, reserving the first K joint nodes with the highest geometric consistency with the candidate skeleton cluster in each camera, and obtaining candidate joint node pairs; S5, using a depth-first search (DFS) to enumerate the cross-camera joint node pairs, and screening optimal joint node pairs in combination with a pruning strategy; and S6, performing triangulation processing on the screened optimal joint node pairs, obtaining three-dimensional skeleton points, and outputting a three-dimensional human skeleton structure.