基于混合高斯模型的运动目标检测  

Moving objects detection based on Gaussian mixture models

作　　者：陈焕钟[1] 李榕[1] 程剑光[1] 

机构地区：[1]华南师范大学,广州510006

出　　处：《激光杂志》2009年第4期32-33,共2页Laser Journal

摘　　要：针对现有方法在复杂的环境下不能很好地检测出运动物体的问题,提出了一种改进的基于混合高斯模型的背景消减法检测运动目标。改进了背景模型的更新算法,提高了背景更新速度。利用帧间差分法消除了运动影子和光照突变问题,并采用(r,g,I)颜色空间检测和抑制了运动阴影。实验结果表明,该算法能很好地提取出运动目标。In the view of the current complex environraent, the detection of moving objects can not be satisfactory with traditional methods. The background of diminishing method based on the improved mixture Gaussian model was presented, a new method is used to update the background more quickly. Through using the frame difference, the problem of moving ghost and quick illumination changes is solved. It also uses the （ r g l ） color space to detect and remove the moving shadows. The experimental results indicate that the moving body can be extracted effectively.

关 键 词：背景模型 背景更新 运动影子 运动阴影 

分 类 号：TN248.1[电子电信—物理电子学]