基于背景差分的一种运动目标检测方法  

A Moving Objects Detection Method Based on Background Subtraction

作　　者：张传伟[1] 王京梅[1] 林晓明[1] 赵文俊[1] 

机构地区：[1]电子科技大学微电子与固体电子学院,四川成都610054

出　　处：《电子科技》2015年第10期69-71,77,共4页Electronic Science and Technology

摘　　要：针对运动目标检测中,传统背景差分法在运动目标和背景颜色相近时不易检测的缺点,提出了一种检测完整运动目标的方法。该方法对YUV彩色空间下的3个通道分别选取独立的阈值进行初次检测,最大化地利用了视频中图像的色彩信息。在包含初次检测所获运动目标的最小矩形区域内进行二次检测,有效地提高了检测精度。实验证明,相比于常规方法,该方法的检测结果更加清晰完整。Abstract Aiming at the failure of traditional background subtraction in the moving detection when the moving object is similar in color with the background, a complete motion objects detection method is proposed. The method selects independent thresholds for the three channels of the YUV color space to do the first detection, making full use of the image color information in the video. A second detection is performed in the smallest rectangle area that contains the first detected moving object to improve the detection accuracy effectively. Experiments demonstrate that the detection results by this method are more clear and complete than those by the traditional method.

关 键 词：运动目标检测 背景差分法 YUV彩色空间 阈值 

分 类 号：N391.41[自然科学总论]