摘要
ViBe算法简单、快速,具有较好的前景检测性能,是运动目标检测和背景建模的主要方法之一。但是在动态背景、相机抖动等户外视频中仍存在噪声和干扰等问题,导致对前景运动目标的检测不准确。针对此问题,提出用像素帧差值代替像素值来初始化背景样本模型的方法,并根据背景动态变化自适应更新阈值来分割前景与背景。实验结果表明,改进算法提高了前景检测的准确性,对噪声干扰表现出了良好的鲁棒性。
Vibe algorithm is simple,fast,and has good foreground detection performance. It is one of the main methods of moving object detection background modeling. But it is still hard to detect the foreground for the outdoor video be cause of the complex background, such as camera shake or trembling leaves of the trees which leads to the inaccurate detection of the moving object. We presented a novel algorithm for moving object detection from a video. The improved approach of ViBe follows a new background model which uses the frame differencing instead of pixel value. Since the fixed threshold value of ViBe algorithm cannot reflect the change of background in real time, a method with self-adaptive threshold was proposed. The experimental results show that the improved algorithm can improve the accuracy of fore- ground detection, and it has good robustness against disturbance.
出处
《计算机科学》
CSCD
北大核心
2017年第9期304-307,共4页
Computer Science
基金
湖北省自然科学基金(2014CFB485)资助
关键词
运动目标检测
背景建模
ViBe算法
帧差
自适应阈值
Moving object detection,Background modeling, ViBe algorithm,Frame differencing, Self-adaptive threshold
