基于帧间差分和粗糙熵的运动目标检测算法  

Moving Target Detection Algorithm Based on Frame Difference and Rough Entropy

作　　者：靳海伟[1] 彭力[1] 卢晓龙[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《江南大学学报（自然科学版）》2015年第1期28-33,共6页Joural of Jiangnan University (Natural Science Edition)　

基　　金：江苏省产学研联合创新项目(BY2014023-25)

摘　　要：帧间差分法只把相邻帧间存在较大差异的部分提取出来,检测出的运动目标内部往往存在"空隙"。利用粗糙熵可以精确地分割图像区域。引入粗糙熵作为帧间差分法的补充,通过颜色灰度信息完成图像形态学重构,并进行形态学滤波和连通性检测,由此克服了检测出运动目标内部的空洞现象,满足实时性,检测效果显著提高。The frame difference method extracts the large different part of the adjacent images,but the extracted target exists ＂ gap＂. Rough entropy segments image region accurately. Introducing rough entropy as a supplement to the frame difference method,completing the image morphology reconstruction with color grayscale information,then processing the image through morphological filtering and connectivity detection,experiments show that the proposed algorithm overcomes internal cavity phenomenon and meets the real-time. Detection effect is significantly improved.

关 键 词：帧间差分 空隙 目标检测 粗糙熵 形态学重构 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]