基于改进的块直方图方法提取背景技术的研究

Based on the Improved Histogram Method to Extract the Background Technology Research

随着多媒体技术的迅速发展,如何对运动物体进行实时准确的检测,成为一个重要的研究课题.本文在分析了背景差分、帧间差法和光流法3类运动检测方法的基础上,针对光流法存在运算复杂、帧间差法实时性差、背景差分提取速度慢等缺点,重点分析了背景差分算法,提出了基于改进的块直方图方法提取背景.利用改进后的方法进行直方图分析时,不是针对固定大小的模块,也不是像素点,而是针对同类目标的类模块.经过仿真分析得出,该方法充分利用同类物体像素之间的相关性,可以减少图像处理计算量,缩短分析时间,同时能得到较好的背景图像.

With the rapid development of multimedia technology, how to detect moving objects in real time accurately has become an important research subject. In this cluding the background difference, frame difference method paper, three kinds of motion detection methods in- and optical flow method are analyzed. Their disadvantages such as complex operation for optical flow method, poor realtime performance for frame difference method and slow extraction speed for background difference are also indicated. Based on this, the background differ- ence algorithm is mainly analyzed, background extraction based on the improved block histogram is proposed. U- sing the improved method of histogram analysis, it is not for fixed size module, nor pixels, but against the class module of the same target. Through the simulation analysis, it can be concluded that the method makes full use of correlation between similar object pixels, and can reduce the amount of calculation, and short the analysis time, at the same time a better background image can be gotten.