改进PIC的背景重构算法与运动目标检测新方法  

Improved PIC algorithm of background reconstruction and a new method of detecting moving object

作　　者：刘丁[1] 杨延西[1] 赵豆[1] 

机构地区：[1]西安理工大学,西安710048

出　　处：《仪器仪表学报》2010年第4期885-891,共7页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(60675048;60805020)资助项目

摘　　要：根据图像序列中出现频率最高的像素均值为背景点的思想,对像素灰度归类(pixel intensity classification,PIC)算法进行改进,通过将所选取的用于重构背景的序列图像像素值进行归一化、量化统计、量化范围拓展,从而重构背景图像,该方法避免了PIC算法中需要人为设定阈值;舍去了较为耗时及复杂的相近灰度区间合并等步骤;对于重构背景与目标图像作差后的二值图像,提出一种新的目标检测方法:粗精两步搜索法,可以精确确定目标物的位置,实现对运动目标的检测。实验结果表明该方法比PIC方法运行时间短、重构的背景噪声点少、粗精搜索后的目标位置准确,是一种快速有效的运动目标检测方法。Based on the idea that the pixel average appearing with high frequency in an image series is the background points, this paper makes some improvement in PIC algorithm. Through normalization, quantitative statistic, quantization range extension of the pixels of the chosen image series used to reconstruct the background, the background image is reconstructed. The method proposed in this paper avoids threshold setting of the PIC algorithm manually and removes some steps, such as combining the approximate gray scopes and etc. , that need plenty of time and are hard to realize through programming. For the binary image that is the difference of the object image and the reconstructed background image, a new method of detecting moving object is presented. A coarse-fine searching method is suggested to decide the object position exactly and complete the moving object detection. Experiment result proves that the running time of the improved method is shorter than that of PIC algorithm, the noise of the reconstructed background is also reduced and the accurate position of the target can be obtained using the coarse-fine searching method. So the proposed method is a real-time and valid method.

关 键 词：背景重构 目标检测 量化统计 粗精搜索 PIC 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]