一种改进的自适应阈值前景提取方法  

Improved approach for extracting moving targets based on self-adaptive threshold

作　　者：胡珂立[1] 赵鲁阳[1] 林振华[1] 王营冠[1] 

机构地区：[1]中国科学院上海微系统与信息技术研究所无线传感网与通信重点实验室,上海200050

出　　处：《计算机应用研究》2011年第12期4726-4729,共4页Application Research of Computers

基　　金：国家重大专项资助项目(2009ZX03006-001-02);中国科学院知识创新项目(Y15YQA1001)

摘　　要：为了尽可能完整地提取前景,针对视频处理中提取的前景区域出现断层导致单个目标出现分裂的问题,提出了一种改进的自适应阈值前景提取方法。该方法是在非模型法产生背景模型的基础上,通过背景差法获取差分图像,利用环境噪声的相对稳定性,建立一张环境噪声分布图,进而滤除差分图像中存在的环境噪点,自适应求取分割阈值,进行图像二值化处理获取前景区域。实验结果表明,该方法能明显改进前景提取效果,最大化提取目标的完整性。In order to solve the problem of target broken during the video processing, this paper proposed an improved approach for extracting moving targets based on self-adaptive threshold. Once obtained the background subtraction image by nonmodeling background model, built a novel environment noise distribution map as the environment noise was relatively stable. With the help of this map, filtered the environment noise in the background subtraction image. It obtained the adaptive threshold by using the optimal image threshold algorithm through the filtered image, and extracted the moving targets finally. Experiment results show that the proposed approach here can apparendy improve the effect of moving targets extraction.

关 键 词：自适应阈值 环境噪声分布图 背景差 前景提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]