复杂环境下的运动人体骨架提取算法

Algorithm of moving human skeleton extraction in complex environment

针对复杂环境下运动人体难以检测及由人体运动自身的复杂性而引起的骨架提取难的问题,提出了一种复杂环境下视频序列中的运动人体骨架提取算法。算法首先利用区域背景建模获取复杂环境下的背景图像,利用最大色差分量结合自适应阈值分割运动人体;然后根据人体测量数据对人体骨架建模,最后利用Kalman滤波跟踪人体关节点,连接关节点生成运动人体骨架。实验结果表明,该算法能准确地提取复杂环境下视频序列中的运动人体骨架,具有低关节位置误差率。

To detect the moving human body and extract skeleton in a complex environment,this paper proposed a novel algorithm of moving human body skeleton extraction. It obtained the background image by constructing region background model in complex environment,and segmented the moving human body by using the maximum color difference and the adaptive threshold,then comstructed the model of human skeleton,and traced the human body joints by using Kalman filtering,through a connection of joints,generated the skeleton of moving human body. Experimental results show that the algorithm can extract the skeleton of moving human body accurately from video sequence with low error rate of joint position in complex environment.