基于混合高斯模型的目标差分自适应背景模型  

Objects Subtraction Adaptive Background Modeling Based on Mixture Gaussian Model

作　　者：刘永福[1] 潘保昌[1] 郑胜林[1] 谭建斌[1] 

机构地区：[1]广东工业大学,广州510006

出　　处：《现代计算机》2010年第1期63-65,78,共4页Modern Computer

摘　　要：提出一种自适应背景更新率的方法。在混合高斯背景模型基础上,通过混合高斯匹配将前后两帧分割出的运动目标作差分,差分后再次采用混合高斯匹配分辨出过去目标区和当前目标区。在背景更新的过程中,加大过去目标区的更新率,让运动目标在出现停滞状态时给背景模型带来的干扰区域得到快速恢复。试验结果表明,该方法有效地提高背景模型的鲁棒性。Proposes an adaptive background update rate method. Based on Mixture Gaussian background model, through Gaussian match, makes the moving objects separated from the before and after frame and objects subtraction, uses Gaussian match once again to distinguish between the target area in the past and the current target area. When updates the background, increases the refresh rate over the target area, so that the interference area caused by the objects at a standstill situation can be rapidly restored. The experimental result indicates that the method is effective to improve the robustness of background model.

关 键 词：混合高斯模型 目标间差分 自适应背景更新率 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]