联合帧间差分和边缘检测的运动目标检测算法  

Interframe difference and edge detection-based moving object detection algorithm

作　　者：王忠华[1] 王超[1] 

机构地区：[1]南昌航空大学信息工程学院,江西南昌330063

出　　处：《南昌大学学报（理科版）》2017年第1期42-46,共5页Journal of Nanchang University(Natural Science)

基　　金：国家自然科学基金资助(61362036);江西省自然科学基金(2016BAB202037)

摘　　要：面对运动目标检测,采用传统的帧间差分法往往造成目标空洞与轮廓丢失的问题,提出一种联合帧间差分和边缘检测的运动目标检测算法。首先计算相邻三帧的两两差分图像,再采用自适应阈值生成二值化图像,经二值化图像与运算提取运动目标区域图像;其次采用改进型Canny算子依次对相邻三帧图像进行边缘检测,依据运动关联法则,获取运动目标边缘图像;最后联合运动目标区域图像和边缘图像,提取运动目标轮廓,再采用形态学方法提取运动目标。实验表明,本算法提高了运动目标的检测性能和保持了运动目标轮廓。For the detection of moving object, the traditional interframe difference method often leads to ob- ject hole and contour loss. The aim of this paper is to propose a moving object detection algorithm based on interframe difference and edge detection. Firstly, the difference images of adjacent three frame images are calculated in pairs, the binary images are generated using adaptive threshold, and the regional image of mov- ing object is extracted by binary image and operation. Secondly the improved Canny operator is used to de- tect the edges of adjacent three frame images successively. In accordance with the moving association law, the moving object edge image is obtained. Finally,the moving object boundary contour is extracted by com- bining the regional image and the edge images,and then the moving object is extracted by using morpholog ical method. Experiments show that our algorithm improves the detection performance of moving object and maintains the boundary contour of moving object.

关 键 词：帧间差分 边缘检测 目标轮廓 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]