基于改进混合高斯模型的运动目标检测算法  

Moving Object Detection Algorithm Based on Improved Mixture Gaussian Model

作　　者：许益成 谭文安[1] 陈丽婷[2] XU Yi-cheng;TAN Wen-an;CHEN Li-ting(School of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China;School of electrical and information, Taizhou Vocational and Technical college, Taizhou 318000, China)

机构地区：[1]南京航空航天大学计算机科学与技术学院,南京210016 [2]台州职业技术学院电信学院,浙江台州318000

出　　处：《控制工程》2018年第4期630-635,共6页Control Engineering of China

基　　金：浙江省科技厅公益性项目(2014C33074);台州市科技计划项目(1202ky13)

摘　　要：为了获得理想的运动目标检测结果，在分析传统混合高斯模型不足的基础上。提出了一种改进混合高斯模型的运动目标检测算法。该算法首先增大更新率以更好的适应环境变化，并引入像素连通区域思想消除噪声，然后利用目标色度的不变性避免“阴影”现象的出现，最后采用仿真对比实验对其性能进行测试。实验结果表明，该算法可以生成可靠的背景图，能够准确、完整检测出运动目标，获得了比对比算法更优的检测结果，可以满足目标跟踪的实际应用要求。In order to obtain the ideal results of moving object detection, on the basis of in-depth analysis of the Gaussian mixture model, this paper put forward a moving object detection algorithm based on improved Gaussian mixture model. First of all, increasing the update rate to better adapt to the environmental changes of Gaussian mixture model, and the pixels connected region thought is introduced effective to eliminate the noise, and then the target color invariance is used to avoid the emergence of the phenomenon of ＂shadow＂, finally, the experiment results show that the proposed moving object detection algorithm can generate reliable background image, it can accurately detect the moving target and the detection results are ＇better than other moving target detection algorithms and can meet the requirements of the real-time target tracking.

关 键 词：高斯混合模型 目标跟踪 学习速率 阴影抑制 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]