基于块直方图分析的视频背景提取方法

Background Extraction Method Based on Block Histogram Analysis for Video Images

在基于像素的直方图分析背景提取的基础上,提出了基于块的直方图分析背景提取方法.该方法在提取背景像素时利用了时域帧间像素的统计特性和帧内局部区域像素的相关性,在直方图分析时不是统计图像中的单个像素点,而是2×2的块,从而显著地减小了利用直方图分析提取背景时的计算量.给出了基于块的方法与普通基于像素的直方图分析背景提取方法的对比试验.试验结果证明了该方法具有提取背景速度快、背景质量好的优点.

A novel method of histogram analysis for background extraction of video images was proposed, which was derived from the pixel-based histogram analysis. The method make full use of the statistical properties of pixels between temporal frames and the correlation of local pixels in a single frame. When carrying out histogram analysis for background extraction, the proposed method is based on a 2 × 2 block of pixels, instead of a single pixel, so that the number of calculations are dramatically reduced, and can extract a sound background image from video sequence, simultaneously. Experiment was presented for comparison between the proposed method and the conventional pixel-based histogram analysis, and the results show that the proposed mehtod is faster and delivers better extracted backgrounds.