摘要
最近,利用步态对个人身份进行识别受到越来越多生物识别技术研究者的重视。步态能量图(Gait EnergyImage,GEI)是一种有效的步态表征方法。把步态能量图分解为身体相关能量图(Body-Related GEI,BGEI)、步态相关能量图(Gait-Related GEI,GGEI)、身体步态相关能量图(Body-Gait-Related GEI,BGGEI)3部分,利用傅立叶描绘子对身体相关能量图(BGEI)、身体步态相关能量图(BGGEI)进行描述,利用Gabor小波提取步态相关能量图(GGEI)的幅值特征,分别研究了它们的识别能力,并在Rank层和Score层融合步态相关能量图(GGEI)、身体步态相关能量图(BGGEI)这两部分信息用于步态识别。该算法在CASIA数据库上进行的试验取得了较高的正确识别率。
Recently,gait recognition for individual identification has received much increased attention from biometrics researchers.Gait Energy Image(GEI) is an efficient represent method.We divided GEI into three parts—Body related GEI(BGEI),Gait related GEI(GGEI),Body Gait related GEI(BGGEI).The fourier descriptor was used to describe the BGEI and BGGEI.The Gabor wavelet was used to the GGEI to get the magnitude feature,research their recognition ability respectively,fusion the two parts of Gait-Related GEI and Body-Gait-Related GEI in rank level and score level to gait recognition.The algorithm was tested in the CASIA datasets and gained high correct recognition rates.
出处
《计算机科学》
CSCD
北大核心
2012年第4期261-264,共4页
Computer Science
基金
国家自然科学基金项目(60975083
31100958
U0835005)资助
关键词
步态能量图
静态特征
动态特征
融合
步态识别
Gait energy image
Static feature
Dynamic feature
Fusion
Gait recognition