基于不变矩特征和神经网络的步态识别

Gait recognition based on moment invariants

步态识别是利用人体步行的方式来识别人的身份。近年来,步态作为一种生物特征识别技术已引起越来越多人们的兴趣。本文提出了一种简单有效的步态识别算法,首先通过背景差方法得到运动人体轮廓,然后利用不变矩描述轮廓特征,最后用BP神经网络方法来进行模板匹配,实现人的身份识别。

Gait recognition is the process of identifying individuals by the walking manner. In the last years, the gait as one of bio- metrics has aroused more and more interest,This paper proposes a simple and efficient gait recognition algorithm, body silhouettes are extracted by background subtraction, and we describe the body silhouettes by moment invariants,the recognition is achieved by BP neural network.