一种改进的三帧差分运动目标检测算法  

An improved three-frame difference algorithm for moving object detection

作　　者：陈宝远[1] 霍智超 陈光毅[1] 孙忠祥[1] 

机构地区：[1]哈尔滨理工大学测控技术与仪器黑龙江省高校实验室,黑龙江哈尔滨150080

出　　处：《应用科技》2016年第2期10-13,共4页Applied Science and Technology

基　　金：黑龙江省教育厅科技项目(12531113);哈尔滨理工大学仪器研究项目(2012XZA04014)

摘　　要：针对传统的三帧差分运动目标检测算法存在的通用性差和检测目标不完整的问题,提出一种改进的三帧差分运动目标检测算法,并给出了试验中的检测效果对比图。该算法采用自适应阈值对差分结果进行二值化,从而提高了算法的通用性。在二值化图像上画出运动目标外部轮廓并对目标轮廓内部进行单色填充,弥补了运动目标内部的空缺。实验结果表明,该算法运算量小,实时性高,通用性好,可以更准确地检测出运动目标,并解决了目标内部空洞的问题。For poor generality and incomplete object detection in the traditional three-frame difference algorithm for moving object detection,an improved three-frame difference algorithm for moving object detection is proposed and the detection effect contrast figures are given. The algorithm adopts adaptive threshold to binarize the difference results,so as to improve generality of the algorithm. The outer contour of moving objects is drawn for the finery image and the internal contour is filled with monochrome color,filling the internal vacancies of the moving objects. The experimental results show that the algorithm has low computational complexity and high real-time performance and good generality. It can detect moving objects more accurately and solve the problem of the internal empty of objects.

关 键 词：三帧差分 运动目标检测 自适应阈值 二值化 内部空洞 

分 类 号：TN911[电子电信—通信与信息系统]