现代电子技术2017，Vol.40Issue(11)：65-67,3.DOI:10.16652/j.issn.1004-373x.2017.11.017
改进的自适应高斯混合模型运动目标检测算法
Moving target detection algorithm for improved adaptive Gaussian mixture model
摘要
Abstract
Since the traditional Gaussian mixture model has fixed learning rate and distribution number,can′t describe the transformation background accurately,and exists the data redundancy,the Gaussian mixture model was improved in the following three aspects. In the stage of the model initialization,the various initial distribution numbers are set according to different envi-ronments. The value of learning rate is adjusted dynamically according to the speed of the environmental change. The Gaussian distribution is updated continuously to delete the model that can′t satisfy the requirement,and create the new distribution. The experimental results show that,in comparison with the traditional Gaussian mixture model,this adaptive Gaussian mixture model can improve the detection accuracy of moving object significantly.关键词
高斯混合模型/运动物体检测/高斯分布/学习率取值Key words
Gaussian mixture model/moving object detection/Gaussian distribution/learning rate valuation分类
信息技术与安全科学引用本文复制引用
唐洪良,黄颖,黄淮,杨成顺,黄宵宁..改进的自适应高斯混合模型运动目标检测算法[J].现代电子技术,2017,40(11):65-67,3.基金项目
南京工程学院引进人才科研启动基金(YKJ201412) （YKJ201412）
