摘要
针对运动目标检测中场景的混乱多变和干扰的复杂多样的问题,提出了一种鲁棒而有效的运动目标检测方法。通过对混合高斯模型的匹配准则和背景模型学习更新方法进行改进,使背景模型的可靠性和收敛速度得到了有效的提高。根据各种干扰的特点,分别实现了光照变化、物体的移入移出的干扰检测和排除。实验结果证明,本文提出运动目标检测算法具有较好的实时性和鲁棒性。
A detection algorithm of moving objects which was robust and effective was proposed for the problem of the chaos and va- riety scenes and the complex and varied interference. The reliability and convergence speed of Gaussian Mixture Model （GMM） were effectively improved with the modified matching criteria and the methods of background learning and updating. According to the char- acteristics of various kinds of interference, the algorithm can detect and remove the interferences with sudden changes in illumination and the movement of object. The experimental results demonstrate that the algorithm has good real-time and robustness for moving object detection.
出处
《微计算机信息》
2010年第4期236-238,共3页
Control & Automation
基金
2008年广西研究生科研创新项目
基金申请人:白雪
项目名称:活动背景的弱小运动目标检测技术研究
基金颁发部门:广西区教育厅(2008105950811M421)
