基于Opencv的复杂交通运动目标检测算法研究  

Research of Video Moving Object Detection in Complex Traffic Based on OpenCV

作　　者：刘馨[1] 张斌 

机构地区：[1]兰州工业学院电子信息工程系 [2]95118部队57分队

出　　处：《自动化与仪器仪表》2013年第4期12-13,16,共3页Automation & Instrumentation

基　　金：甘肃省自然科学基金项目(096RJZA084);甘肃省科技支撑项目(1104GKCA032);兰州市科技计划项目(2010-1-225)

摘　　要：城市复杂交通场景中环境因素不可预测,针对运动目标的自动检测问题,提出一种改进的背景差分算法,采用背景图像多帧平均进行背景的获取和更新,利用改进的OSTU阈值进行差分图像分割并提取目标特征。在Opencv环境下,应用该算法对复杂交通场景下的运动目标车辆进行检测。实验结果表明,该算法能正确地检测出运动车辆,具有一定的有效性和可行性。It is not predict for urban environmental factors in complex traffic. This paper presents an improved algorithm based on background subtraction for automatic detection of moving targets, which background image can be obtained and updated by a multi-frame average, image segmentation and feature extraction of the target can be gotten by OSTU threshold. It can be used to solve the problem of traffic vehicle detection in the complex traffic scenarios with OpenCV library function. The results show the improved algorithm can detect moving vehicles correctly, so it has a certain validity and feasibility.

关 键 词：背景差分法 OPENCV 基于背景图像多帧平均 OSTU阈值算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]