摘要
介绍一种结合背景差分法和瞬时差分法的在视频序列中识别运动目标算法。该算法利用瞬时差分法得到当前帧中运动目标的轮廓信息,在更新背景模型时不更新运动目标轮廓内区域,避免了由运动目标引起的背景模型更新误差。给出用OpenCV实现算法的具体过程和关键代码。实验结果表明,该方法可以实现视频序列中运动目标的识别,具有实时性,并能得到较好的识别结果。
an object recognition algorithm which integrates the background subtraction algorithm and the temporal differencing algorithm is introdued. This algorithm obtains outline of the current frame moving object by temporal differencing algorithm,and then updates background without the information inside the outline. This algorithm can avoid updating error by moving object. It also presents how to implement the algorithm with OpenCV in detail,and lists the pivotal codes of the program. The experiment shows that this algorithm can recognize moving object in the video files in real - time.
出处
《现代电子技术》
2009年第22期99-101,共3页
Modern Electronics Technique
