基于视觉特征的人形定位算法的提出及实现  

Realization and Proposition of a Algorithm for Human Shape Location Based on Vision Features

作　　者：毛志智[1] 曹丽武[1] 吴杰[1] 

机构地区：[1]华中科技大学控制系,武汉4300074

出　　处：《计算机与数字工程》2010年第10期109-112,共4页Computer & Digital Engineering

摘　　要：文章针对嵌入式红外安防系统中红外误报的问题,研究和提出将人形侦测引入到嵌入式安放系统中的来的观点。针对人形检测和定位的问题,提出了一种基于视觉特征的人体头肩模型,依据此模型给出了在基于ARM7微控制器环境下,人形识别的硬件处理方案,运动区域的提取方法和一种基于最大面积策略的人体运动区域的选择方案。最后利用MATLAB仿真环境和嵌入式系统的实际仿真,得知算法具有一定的针对性和有效性,能够解决直立人环境下人形的定位和检测问题。In this paper,the idea to introduce human shape detection to embedded system to deal with the problem of infraied fault.At the same time,a model of human body head and shoulder using vision features was proposed to complete human shape detection and location.A hardware method in the ARM7 environment and the area selection method based on maximum area were put up.In the end an experiment was made under MATLAB emulation environment and the real embedded system.The result showed that this algorithm is relatively right and it can solve human shape detection and location problem in straight human environment.

关 键 词：人形检测 视觉特征 红外误报 ARM7 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]