摘要
本文对交通视频中广泛应用的背景差分法进行了研究,在传统视频图像背景差分模型的基础上分析了背景噪声与运动目标灰度值的分布特征,针对采用单阈值分割的传统背景差分法存在的一些不足,本文提出了一种更为合理的背景差分模型:通过分别对背景差分图像明、暗两侧分开处理,同时利用Otsu(最大类间方差)算法找出左右两个最佳阈值来实现更优的视频图像的分割。
The paper has studied widespread-application of background difference method in traffic video,and analyzed the distribu-tion characteristics of the background noise and the moving target gray value with traditional background of the video image differencemodel. As the traditional single-threshold segmentation of the background difference method exists some shortcomings,this paper pro-posed a more rational background difference model：by processing each part of light and dark respectively and using the Otsu algo-rithm to find out two optimal thresholds,to achieve a better video image segmentation.
出处
《微计算机信息》
2010年第29期206-207,180,共3页
Control & Automation
关键词
背景模型
图像分割
车辆检测
Background model
Image segmentation
Vehicle detection
