【作者】 王贵珠； 李俊韬； 范耀祖；

【Author】 Wang Guizhu Li Juntao Fan Yaozu (School of Automation Science and Electrical Engineering, Beihang University, Beijing, 100083, China)

【机构】 北京航空航天大学自动化科学与电气工程学院；

【摘要】 运动目标检测的主要目的是在视频序列中分割出具有意义运动对象实体,减背景法能够很好地从一段视频中提取出运动目标。本文提出一种基于阴影抑制和统计背景模型的运动目标检测算法。该算法首先采用一种统计的方法提取初始背景,并根据卡尔曼滤波理论对背景进行更新;然后采用一种自适应阈值法进行运动目标检测;后处理后,通过度量当前图像与背景图像除运算后图像的平坦度方法来消除阴影。实验结果证明,本文算法效果很好,能准确地检测运动目标并去除阴影。更多还原

【Abstract】 The objective of moving objects detection is segmenting and extracting representative body from the images. The method of making difference between the current frame and the background can extract moving objects from a video. This paper presents a method of moving objects detection based on shadow suppression and statistical background model. Firstly, a statistical method is used to obtain initial background image and Kalman filtering theory is used to update the background image ; Then , detect the moving objects use adaptive threshold.; After post treatments, through measuring the flatness degree of the result of image division to eliminate shadows. Experimental results show this method perform well and can exactly detect moving objects without shadow.更多还原