基于动态二维图像序列的三维步态识别方法.pdfVIP

第29卷第 10期 Vol.29 No.10 2013年10月 BULLETINOFSCIENCEANDTECHNOLOGY Oct. 2013 基于动态二维图像序列的三维步态识别方法 王运成 （聊城大学东昌学院 ，山东聊城 252000） 摘 要：针对在当前二维步态识别系统中，受到人体走路姿势中的特征重叠的影响，不能很好的解决信 息丢失问题。而三维识别也面临着技术要求复杂，对设备角度要求高的问题，本文提出了一种基于动态 二维图像序列的三维步态识别方法。利用摄像机采集单帧步态图像序列，通过对单帧连续通信进行坐 标动态转换，对特征矩阵进行有效的三维转化，计算三维特征数据，提取三维人体单帧二维图像转化的 特征，进行步态识别。结果表明，本方法无需复杂的设备，能够解决特征丢失问题，识别准确度有了较好 的改善。 关键词：二维图像；三维步态识别；特征转换 中图分类号：TP391.9 文献标识码：A 文章编号：1001-7119（2013）10-0186-03 Three-Dimensional GaitRecognitionMethodBased onDynamic2D Image Sequence WangYuncheng (DongchangCollegeof Liaocheng University,ShandongLiaocheng252000,China) Abstract：Based on the current two-dimensional gait recognition system, the human walking by features in the overlap原 ping influence, not a very good solution to solve information loss problem. And three dimensional recognition also faces the complex technical requirements, the equipment Angle demanding problem, this paper proposes a dynamic image se原 quence based on two-dimensional three-dimensional gait recognition method. The use of video cameras to acquisition single frame gait image sequence, through to the single frame continuous communication to coordinate dynamic conver原 sion, the characteristic matrix to effectively 3 d transformation, calculation of three-dimensional features data, extracting 3 d body single frame 2 d images into the characteristics of gait recognition. The results show that the present method this method doesnt complicated equipment, ableto solvefeature loss problem, recognition accuracy hav