Estimation Of Human Motion Parameters Based On Feature Point Extraction Technology

With the rapidly developing of the image and video technologies,numbers of human motion image or video were provided to study the human motion.Human motion parameter estimation is the most important and most popular research direction.Compared with the general color picture,the 3D image can be obtained by the relevant depth camera,which provides more information than the general picture.Therefore,this thesis will combine the above two research directions to study the algorithm for extracting the feature points of the 3D image and how to An algorithm that matches feature points in different data.And the human motion parameter can be estimated based on the above algorithm.This thesis studies the feature point extraction of point cloud images.Based on some simple methods,this paper draws on the characteristics of implicit surfaces and B-spline functions,and proposes an implicit surface method based on B-spline function to fit the complete human body from discrete point cloud data.The surface,and then from the surface of the human body to obtain the feature points characterizing the human body.Similar methods have been used to recover complete original images from two-dimensional photos of partial data.This paper extends the relevant methods to the 3D point cloud,which enables us to extract human feature points more accurately from the messy 3D point cloud.In addition,this thesis also studies how to use the feature points to obtain motion parameters.The human motion equation is constructed by using feature points,and the correlation least squares method is used to calculate,and a method for easily expressing motion parameters by quaternion is given.