基于模糊理论决策的双向二维PCA步态识别算法.doc

图 学 学 报 JOURNAL OF GRAPHICS 2012 年 12 月 第 33 卷 第 6 期 December 2012 Vol.33 No.6 基于模糊理论决策的双向二维 识别算法 PCA 步态 陈祥涛1， 张前进2， 张双玲3 （1. 河南科技大学现代教育技术与信息中心，河南 洛阳 471003； 2. 河南科技大学电子信息工程学院，河南 洛阳 471003； 3. 河南省工业设计学校机电工程系，河南 郑州 450002） 摘 要：针对步态识别中的平均步态能量图像系数矩阵维数过高和分类较困难的特 点，提出一种基于模糊理论决策分类的双向二维主成分分析的步态识别算法。通过预处理技 术得到平均步态能量图并将得到的图像分割为多个子图像，利用双向二维主成分分析来降低 平均步态能量子图像的系数矩阵维数，加快识别速度。引入模糊理论决策的方法进行最近邻 分类器的分类。最后在 CASIA 步态数据库上对所提出的算法进行实验，实验结果表明该算 法具有较好的识别性能并有较强的鲁棒性。 关 键 词：步态识别；平均步态能量图；主成分分析；双向二维主成分分析 中图分类号：TP 391 文献标识码：A 文 章 编 号：2095-302X (2012)06-0103-07 A gait recognition algorithm using fuzzy theory decision double-directional two-dimensional principal component analysis Chen Xiangtao1, Zhang Qianjin2, Zhang Shuangling3 ( 1. Modern Education Technology and Information Center, Henan University of Science and Technology, Luoyang Henan 471003, China; 2. School of Electronics and Information Engineering, Henan University of Science and Technology, Luoyang Henan 471003,China; 3. Electrical Engineering Design Department, School of Henan Indutrial Design, Zhengzhou Henan 450002, China ) Abstract: A gait recognition algorithm is proposed based on fuzzy theory classification decision double-directional two-dimensional principal component analysis (DTPCA) for the problems of high dimensionality and difficult classification of mean gait energy image (MGEI) coefficient matrix. A preprocess technique is used to obtain MGEI, which is then divided into some of sub-image blocks, then DTPCA is used to reduce the dimension of mean gait energy sub-image blocks coefficient matrix and improve the recognition speed. The fuzzy theory is introduced to classify the nearest-neighbour classifiers. The proposed algorithm is evaluated on CASIA Gait Database. Experimental results show that the proposed approach achieves a high recognition accuracy and stronger robustness. Key words: gait recognition; mean gait energy image (MGEI); principal component analysis (PCA); double-directional two-dimensional principal com