中文摘要：

针对传统LBP纹理检测在运动目标提取中对原地或缓慢运动物体容易误判为背景的问题,提出一种基于颜色分割和LBP纹理检测的提取方法。主要思想分为3步：根据LBP纹理检测结果映射得到宏块级粗糙运动目标;根据K-means颜色分割方法得到颜色分类信息;根据提出的宏块交叠机制,将颜色信息与运动信息进行融合,得到最终的提取结果。对比实验结果表明,该方法在保持良好的光照鲁棒性和阴影抵抗力的同时,可以有效改善运动目标空洞、背景融入等问题,满足实时要求。

英文摘要：

To solve the problem that situ or slowly moving object is easy to be misjudged as the background using traditional LBP texture detection,a method of moving object extraction based on color segmentation and LBP texture detection was proposed.The main idea consists of three steps：firstly,according to the LBP texture detection results,macroblock level rough moving target was obtained.Secondly,according to K-means color segmentation method,color classification information was obtained.Thirdly,according to the overlapping mechanism of macroblock,the color information and motion information was combined to generate the final results.Experimental results show,the method has good illumination robustness and shadow resistance,and it can effectively improve the moving object extraction effect and meet the real-time requirement.