基于背景差分和三帧差分的运动目标检测

为了提高运动目标检测算法的准确性和对背景变化的适应性,本文采用了三帧差分与基于单高斯模型背景差分法相结合的算法,并通过最大类间方差法提取自适应阈值。引入一个新的背景更新机制,当运动物体融入背景或者背景中物体移除时,将背景更新为当前视频帧。实验结果表明,本文算法在对运动目标进行检测时,不易受背景光线变化及运动物体融入背景等因素的影响,适用于无人监控环境。

Moving Objects Detection Based on Background Difference and Three Frame Difference

To improve the accuracy of moving target detection algorithm and the adaptability to the change of background,this paper combines three frame difference with background subtraction algorithm,which based on single gaussian model,and uses the otsu algorithm to extract the adaptive threshold.A new background update mechanism is introduced,when integrating moving object in the background or removing object in the background,the background update for the current video frame.The experimental results show that the algorithm for detecting a moving target is not easily influenced when background light change or moving objects blend into the background or other factors,and can be applied to unmanned monitoring environment.