摘要
步态作为一个新兴的生物特征,具有广泛的应用前景。现阶段比较成熟的非模型化方法步态能量图,能将一个步态序列表示为单幅的灰度图像,对噪声有较好的鲁棒性和较好的识别效率,但是不能很好的适应人行走速度的改变。因此,本文提出了一种基于关键帧能量固定的步态识别方法。该方法在步态能量图的基础上,对步态序列的关键帧进行了能量固定,将步态能量图转变成为能量固定后的步态能量图再进行特征的提取和识别。实验结果表明,该方法相对传统的步态能量图,能更好的适应速度对识别的影响。
Gait has a wildly using prospect as a new biometric feather.A model-free gait recognition approach named Gait Energy Image(GEI) which represents human motion sequence in a single gray image,was proved less sensitive to silhouette noise and could work better than some other approaches.However,GEI can’t treat the walk-speed problem well.Therefore,this paper propose a new approach based on holding the energy proportion of key-frames immutable in the GEI named KeyframeGEI(KGEI).Our approach based on the proportion of energy in GEI,fi rst we keep the proportion of key-frames’ energy immutable in GEI.Then we use KGEI instead of GEI in the recogniton process.The experiment results show that our approach perform a better performance on the walk-speed problem than GEI.
出处
《信息与电脑(理论版)》
2010年第1期62-64,67,共4页
China Computer & Communication
关键词
步态识别
关键帧
步态能量图
Gait recognition
Key frame
Gait Energy Image