随机体育运动图像捕捉点虚拟重建中凌乱误差控制  

Virtual Reconstruction Error Control of Random Sports Image Capture Point

作　　者：李腾达[1] 赵晓炜[1] 

机构地区：[1]南昌师范学院体育系,江西南昌330032

出　　处：《科技通报》2014年第2期31-34,共4页Bulletin of Science and Technology

摘　　要：运动员进行不同的运动项目时,关节所成的角度,运动速度有差异,造成特征捕捉点不能很好对应关键区域,形成特征凌乱对应误差,影响形状基计算的准确性。传统的体育运动图像三维重建算法,受到形状基凌乱的影响,后期建模准确性很难控制。因此提出随机体育运动图像捕捉点虚拟重建中凌乱误差控制算法。利用因式分解法,建立人体运动时的三维动态仿真模型,运用stolt变换将所有图像捕获点的的方位偏移率调整为相同,使其在进行特征匹配时不会产生过大误差,对于过大的凌乱残余误差,进行三阶运动补偿,实现人体运动图像序列中的三维动态仿真建模。实验结果表明,这种算法提高了人体运动图像序列三维动态仿真的真实性。Athletes with different sports, joint into the point of view, there are differences between the speed and feature points not well corresponding to key areas, forming characteristics of corresponding error, affect the accuracy of shape matrix calculation. The traditional sports image 3D reconstruction algorithm, under the influence of shape base messy, late modeling accuracy is hard to control. Puts forward a random sports image capture point virtual reconstruction error control algorithm. Using the method of factoring, the establishment of three-dimensional dynamic simulation model of human motion, using stolt transform will all image capture points of azimuth migration rate is adjusted for the same, won＇t produce too much when in the feature matching error, residual error for too messy, third-order motion compensation, the 3D dynamic simulation of human motion image sequence modeling. The experimental results show that this algorithm im-proves the human motion image sequence the authenticity of the 3D dynamic simulation.

关 键 词：体育运动 凌乱匹配 形状基 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]