一种改进的自适应阈值前景提取方法

摘要

In order to solve the problem of target broken during the video processing, this paper proposed an improved ap-proach for extracting moving targets based on self-adaptive threshold. Once obtained the background subtraction image by non-modeling 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 thre-shold by using the optimal image threshold algorithm through the filtered image, and extracted the moving targets finally. Ex-periment results show that the proposed approach here can apparently improve the effect of moving targets extraction.%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.