摘要
视频检测技术是智能交通系统研究中一个重要研究方向,根据交通流视频检测的特点,对基于混合高斯分布模型的自适应背景消除方法进行了改进.包括:背景模型匹配只使用亮度信息;将高斯分布模型按权值、方差排序;使用单目深度信息来确定背景;动态调整采样频度等.实验表明,本文提出的算法,分割效果较佳,分割的实时性大大增强.
The video vehicle detection technique is an important research field of ITS. Based on the features of video vehicle detection, this paper improves the method of background subtraction using adaptive Gaussian mixture models. These improvements include: background model matching uses only the luminance information; ranking Gaussian mixture models according to weight and variance; selecting background model using monocular depth information; making dynamic adjustment of sampling rate. The experiments show that the segment quality is good, and real_time performance is greatly improved.
出处
《北方交通大学学报》
CSCD
北大核心
2003年第6期22-25,共4页
Journal of Northern Jiaotong University
基金
高等学校博士学科点专项科学基金资助项目(2000000413)
关键词
图像处理
混合高斯分布模型
背景消除
视频检测
image processing
Gaussian mixture model
background subtraction
video vehicle detection
