Three-Dimensional Diffusion Model in Sports Dance Video Human Skeleton Detection and Extraction

The research in this paper mainly includes as follows: for the principle of action recognition based on the 3D diffusion model convolutional neural network, the whole detection process is carried out from fine to coarse using a bottom-up approach; for the human skeleton detection accuracy, a multibranch multistage cascaded CNN structure is proposed, and this network structure enables the model to learn the relationship between the joints of the human body from the original image and effectively predict the occluded parts, allowing simultaneous prediction of skeleton point positions and skeleton point association information on the one hand, and refinement of the detection results in an iterative manner on the other. For the combination problem of discrete skeleton points, it is proposed to take the limb parts formed between skeleton points as information carriers, construct the skeleton point association information model using vector field, and consider it as a feature, to obtain the relationship between different skeleton points by using the detection method. It is pointed out that the reorganization problem of discrete skeleton points in multiperson scenes is an NP-Hard problem, which can be simplified by decomposing it into a set of subproblems of bipartite graph matching, thus proposing a matching algorithm for discrete skeleton points and optimizing it for the skeleton dislocation and algorithm problems of human occlusion. Compared with traditional two-dimensional images, audio, video, and other multimedia data, the 3D diffusion model data describe the 3D geometric morphological information of the target scene and are not affected by lighting changes, rotation, and scale transformation of the target and thus can describe the realistic scene more comprehensively and realistically. With the continuous updating of diffusion model acquisition equipment, the rapid development of 3D reconstruction technology, and the continuous enhancement of computing power, the research on the application of 3D diffusion model in the detection and extraction of a human skeleton in sports dance videos has become a hot direction in the field of computer vision and computer graphics. Among them, the feature detection description and model alignment of 3D nonrigid models are a fundamental problem with very important research value and significance and challenging at the same time, which has received wide attention from the academic community.

Research on the Detection Method of Implicit Self Symmetry in a High-Level Semantic Model

In order to improve the accuracy of semantic model intrinsic detection, a skeleton-based high-level semantic model intrinsic self-symmetry detection method is proposed. The semantic analysis of the model set is realized by the uniform segmentation of the model within the same style, the component correspondence of the model between different styles, and the shape content clustering. Based on the results of clustering analysis, for a given three-dimensional (3D) point cloud model, according to the curve skeleton, the skeleton point pairs reflecting the symmetry between the model surface points are obtained by the election method, and the symmetry is extended to the model surface vertices according to these skeleton point pairs. With the help of skeleton, the symmetry of the point cloud model is obtained, and then the symmetry region of point cloud model is obtained by the symmetric correspondence matrix and spectrum method, so as to realize the intrinsic symmetry detection of the model. The experimental results show that the proposed method has the advantages of less time, high accuracy, and high reliability.

GEOREFERENCING UAS DERIVATIVES THROUGH POINT CLOUD REGISTRATION WITH ARCHIVED LIDAR DATASETS

Georeferencing gathered images is a common step before performing spatial analysis and other processes on acquired datasets using unmanned aerial systems (UAS). Methods of applying spatial information to aerial images or their derivatives is through onboard GPS (Global Positioning Systems) geotagging, or through tying of models through GCPs (Ground Control Points) acquired in the field. Currently, UAS (Unmanned Aerial System) derivatives are limited to meter-levels of accuracy when their generation is unaided with points of known position on the ground. The use of ground control points established using survey-grade GPS or GNSS receivers can greatly reduce model errors to centimeter levels. However, this comes with additional costs not only with instrument acquisition and survey operations, but also in actual time spent in the field. This study uses a workflow for cloud-based post-processing of UAS data in combination with already existing LiDAR data. The georeferencing of the UAV point cloud is executed using the Iterative Closest Point algorithm (ICP). It is applied through the open-source CloudCompare software (Girardeau-Montaut, 2006) on a ‘skeleton point cloud’. This skeleton point cloud consists of manually extracted features consistent on both LiDAR and UAV data. For this cloud, roads and buildings with minimal deviations given their differing dates of acquisition are considered consistent. Transformation parameters are computed for the skeleton cloud which could then be applied to the whole UAS dataset. In addition, a separate cloud consisting of non-vegetation features automatically derived using CANUPO classification algorithm (Brodu and Lague, 2012) was used to generate a separate set of parameters. Ground survey is done to validate the transformed cloud. An RMSE value of around 16 centimeters was found when comparing validation data to the models georeferenced using the CANUPO cloud and the manual skeleton cloud. Cloud-to-cloud distance computations of CANUPO and manual skeleton clouds were obtained with values for both equal to around 0.67 meters at 1.73 standard deviation.

Curve-Skeleton Extraction Using Appropriate Threshold Optimization

The curve-skeleton of an object is an important abstract geometrical and topological representation of its shape, which is extremely useful for pattern recognition and computer vision applications involving in shape analysis. In this paper, we propose an effective algorithm for extracting curve skeleton based on the definition and properties of curve skeleton from pixel cloud, which integrates the advantages of the visual main parts reliability for object recognition and the skeletons reduced-dimension for object representation. This algorithm can detect each pixel of the image, and find the salience value of each pixel; the salience value is the possibility of the pixel being a skeleton point. Then an appropriate threshold is selected to pruning the skeleton and to get the curve skeleton. In this way, the algorithm can be effective in reducing the number of non-skeleton pixels, and reduce the overall time of extracting skeleton. The experiments show that the skeleton keeps the topology of the target. And the corners of the skeleton are smoother and more natural. In additionally, it can effectively reduce redundant branches of skeleton.

Assessing Coronary Arterial Dynamics from Cineangiograms

Assessing displacement fields of coronary arterial skeletons from a pair of nearly orthogonal cineangiographic image sequences is addressed. Projections of vessel skeletons are firstly extracted from each frame. Then, 2D displacement vectors of vessel skeletons during each temporal interval are estimated along the image sequences. 3D displacement vectors of each skeleton point are finally reconstructed along the overall sequences. Possible errors are discussed. Experimental results with clinically acquired in vivo image data have demonstrated the validity of the method.

The Algorithm to Measure Crack Width with Incircle

Crack is a major defect of buildings. Digital image methods are often used to detect cracks. But incorrect or un-unique results may be inverted with an inappropriate algorithm. An image processing way is presented to obtain the sole width value. Meanwhile, the crack with several branches can be measured. In the processing, the crack skeleton is first calculated. Then each of the points on the skeleton is served as a center of a group of circles, one by one. The radius of the circles is increased step by step. The iterations will not stop until any point in the circle goes out of the crack. Thus the last circle in the iteration is served as an incircle of the crack. The diameter of the incircle is a crack width in a given skeleton point. The maximal and average width of the crack will be calculated after all the incircles with all the skeleton point are traversed. The experimental results show the proposed method can extract the width of cracks in a complex context.