【摘要】 针对现有步态识别算法在步态周期变化时不能很好地保留瞬时信息的问题,提出了一种基于傅里叶描述符优化形变轮廓插值方法。首先,对人类轮廓的周期性形变进行建模;然后,将基于傅里叶系数乘积的新度量作为闭合曲线之间的距离度量,并利用帧插值方法近似识别半周期的起始帧和结束帧;最后,利用步态距离完成最终的步态分类。在OU-ISIR和CASIA步态数据库上的识别精度可分别高达99%、87.34%,分析结果表明,提出的算法在步态周期变化时依然能保留瞬时信息,对行走速度上的轻微变化具有更好的鲁棒性,并且大大降低了整体计算复杂度。更多还原

【Abstract】 Existing Gait recognition algorithms can not save moment information well when gait cycle varying,so an interpolated deformable contours algorithm optimized by Fourier descriptor is proposed. Firstly,the model of the periodic deformation of human contours was built. Then a new measure of similarity using the product of Fourier coefficients was proposed as a distance measure between closed curves,and the frame interpolation method was used to approximate recognize the start frame and end frame of half cycle. Finally,the final gait classification was completed through gait distance. Recognition accuracies on OU-ISIR gait database and CASIA gait database can achieve 99% and 87. 34% respectively. Experimental results show that the proposed approach is able to preserve temporal information even when the gait cycles change,and therefore offers a better robustness to slight variation in walking speed. Besides,the approach proposed can reduce the overall computational complexity in actual implementation evidently.更多还原