步态识别论文，对目标检测方法进行了分析，提出了在HSL颜色模型空间中，利用时间域中值滤波算法构建背景模型，采用背景减除法实现人体上肢和下肢关节点的检测，采用闽值分割、形态学滤波和颗粒去除操作对关节点的图像进行二值化处理，为后续相关特征的提取做好了准备。

-Gait identification papers, for target detection methods are analyzed, presented at the HSL color space model, using median filtering algorithm in the time domain to build the background model, to achieve detection of human upper limb and lower limb joints using background subtraction method, using threshold segmentation, morphological filtering and particle removal operation for image joints were binary processing, ready for subsequent extraction of relevant features.