【作者】 王彪； 王成儒； 王芬芬；

【Author】 WANG Biao1 WANG Cheng-ru1 WANG Fen-fen21 College of Information Science and Engineering, Yansan university, Qinhuangdao 066004, China 2 Department of Automation Northeast University at Qinhuangdao, Qinhuangdao 066004, China

【机构】 燕山大学信息科学与工程学院； 东北大学秦皇岛分校自动化工程系；

【摘要】 提出了一种运动目标检测方法。用混合高斯模型得到运动目标,并结合像素的时间与空间特性,提出区域选择更新混合高斯模型来抽取背景和使用某像素点属于背景和前景的概率大小来进行分类。当某个前景目标的像素被错误的分类为背景,使用该方法就可以避免错用前景目标像素颜色去更新背景模型,因而具有更好的检测效果。该方法还能检测与背景区别不明显的前景目标,具有较快计算速度,满足实时系统的要求。更多还原

【Abstract】 An approach to detecting a moving object presented. Moving objects are segmented by using hybrid Gaussian model as background. To take into account temporal and spatial relation on pixels,it presents Region Selective Update Mixture Gaussian Background Mode. When a pixel should be foreground, but the color is still described well enough by the background model to be classified as background, the background model updated with foreground colors, which leads to a bad repression of the real background and subsequently bad classification results. Using that method can solve this kind of circumstance and have a good segmented effect. This method can detect moving object when the background and foreground color distributions are not distinct, and compute fast enough to suffice real-time system.更多还原