一种基于边缘特征的背景差分运动目标检测方法  

A Background Subtraction Motion Object Detection Method Based on Edge Feature

作　　者：董久敏[1] 王枚[1] 潘国华[2] 宋巍巍[3] 

机构地区：[1]烟台职业学院信息工程系．山东烟台264670 [2]烟台市公安局,山东烟台264000 [3]92323部队,山东烟台264000

出　　处：《海军航空工程学院学报》2011年第2期173-176,共4页Journal of Naval Aeronautical and Astronautical University

摘　　要：针对背景差分运动目标检测中目标边缘不够精确的问题,提出一种基于边缘特征的背景差分运动目标检测方法。该方法结合了均值背景的总体特性和对称帧差的实时性。首先,利用包含同一图像的两幅对称帧差相与结果获取运动目标的准确边缘,并将其作为目标分割的边界限制;然后,利用背景差分方法获取运动目标区域范围,提取目标像素点作为区域生长的种子点;最后,利用区域中种子点的颜色特征在边界限制范围内准确分割目标。实测彩色图像试验结果表明,与典型方法相比,基于边缘特征的目标检测方法可以快速获得更为准确的目标分割结果。For the problem of object imprecise detection of background subtraction method,the object detection method based on edge feature was proposed.The method merges the integrality of mean-value background subtraction and real-time feature of symmetry frame subtraction.Firstly,the object edge was extracted by symmetry frame subtraction of two images with the same middle image,and the object edge was regarded as the limitation ruler of region growing.Secondly,the motion object position was given by background subtraction.At last,the precise motion object position was detected by region growing based on object edge.Lots of actual vehicle images were used to test;the experimental result demonstrated the accuracy and effectiveness of the proposed algorithm.

关 键 词：边缘特征 背景差分 对称帧差 运动目标检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]