摘要
自动、准确和快速检测人体头部是当今计算机视觉研究中的一项关键任务。本文提出一种基于背景差分和改进组合模板的人体头部精确定位方法。利用背景差分法得到人体的大概轮廓,然后利用人体头部与躯干相对关系初步确定人体的头部位置,再运用模糊组合模板法得到头部的精确定位。实验结果表明:该方法能有效地提高人体头部定位算法的速度与精确度。
It is a key task that automatic, accurate and rapid head detection in the computer vision.This paper presents a novel head-location method, based on background difference picture and combined mask.Firstly, the method gets the general outline of the human body by background difference picture.Secondly, it adjusts the coarse position to the accurate position by adopting two novel masks.Finally, it receives a correct detection by fuzzy clustering method.The experimental results show that the proposed method can advance the speed and precision of the head location.
出处
《微计算机信息》
2010年第8期197-199,共3页
Control & Automation
关键词
人体头部提取
背景差分法
组合模板
模糊聚类
Head-location
Background Difference Picture
Combined Mask
Fuzzy Clustering
