摘要
为了实时准确地重建出背景,提出了一种基于帧间差分的背景重建算法。该算法等间隔采样视频帧,然后对视频序列进行帧间差分,对得到的差分图像分块处理,通过比较各子块的亮度与能量均值,将各帧中满足要求的子块进行组合,从而快速地重建出背景。当背景发生整体或局部变化时,该算法能够快速地检测出背景变化,并采用相应算法实时更新背景。实验结果表明,该算法能快速、准确地重建出背景,从而能够完整地提取场景中的运动目标。
In order to reconstruct a background in real time and correctly, a background reconstruction algorithm based on differ- ence between frames was proposed. In this algorithm, the video frames were picked up every few frames, then the retrived differ- enee image of the adjacent frames was calculated, the difference image was divided into some sub-regions, with the average inten- sity and image energy as the constraints, the blocks which met the requirements of background were assembled together, so the background could be reconstructed fast in this way. When the background changed completely or partially, this algorithm could detect and update the change in real time with corresponding algorithm. Experimental results indicate that the background can be reconstructed fast and correctly in the proposed algorithm, so the moving objects can be extracted completely and successfully.
出处
《机电工程》
CAS
2008年第8期11-14,17,共5页
Journal of Mechanical & Electrical Engineering
基金
浙江省科技厅计划资助项目(2007C21049)
关键词
减背景法
背景重建与更新
帧间差分
亮度与能量均值
图像分块
background subtraction
background reconstruction and update
difference between frames
average intensity and image energy
image division
