基于自适应高斯模型的实效运动目标检测算法  

Efficient moving target detection method based on adaptive Gaussian model

作　　者：王红茹[1] 童伟[1] 

机构地区：[1]江苏科技大学机械工程学院,江苏镇江212003

出　　处：《计算机工程与设计》2016年第10期2700-2704,共5页Computer Engineering and Design

基　　金：江苏科技大学高级人才基金项目(35020902)

摘　　要：针对复杂环境下经典混合高斯背景建模算法实时性差以及帧间差分法检测精度低的问题,提出一种基于七帧差分和改进的自适应混合高斯模型相结合的运动目标检测算法。通过七帧差分获取当前帧运动目标的粗略区域;利用HSV颜色空间色度的不变性进行阴影抑制,提取出背景区域和可疑运动区域;对可疑运动区域使用混合高斯法区分出背景显露区域以及运动区域,对每个区域的高斯建模参数采用不同的更新策略,不对背景区域进行高斯匹配;引入光照突变参数,若发生光照突变,对高斯模型中的建模参数重新初始化。对比实验结果表明,该算法能有效抑制阴影和光照突变对检测精度的影响,具有良好的实时性。To cope with poor real-time performance of traditional Gaussian mixture background model method and low detection accuracy of frame difference method in complex environments,a moving target detection method based on seven frame difference and improved adaptive Gaussian mixture model was proposed.Rough region of the current frame was extracted using seven frame difference.The shadow was suppressed using the invariability of HSV color space chromaticity to get background region and suspicious motion region subsequently.Suspicious motion region was divided into the exposed background region and motion region using mixed Gauss algorithm.Mixed Gaussian parameter of each region was updated according to different strategies and the background region was not matched by Gaussian.Light mutation parameter was introduced to judge light mutation and Gauss model parameter was initialized.Comparative experimental results indicate that the proposed method not only can effectively restrain the influence of shadow and light mutation on the detection accuracy,but also has good real-time performance.

关 键 词：运动目标检测 七帧差分 自适应更新 混合高斯模型 颜色空间 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]