人体行为检测与识别

摘要

人体行为检测与识别是当前研究的重点，具有很高的研究价值和广阔的应用前景。主要应用在型人机交互、运动分析、智能监控和虚拟现实也称灵境技术（VR)领域，对于研究人体检测和识别有着重要的意义。因为人的运动的复杂性和对外部环境的多变性，使得人们行为识别和检测具有一些挑战。对人类行为和检测的研究目前处于初级阶段，有待进一步研究和开发。

本文基于matlab人体行为识别和检测的研究，本文主要研究的是从图像中判断出目标处于何种行为，例如常见的走、坐、躺、跑、跳等行为的识别。从现有的很多主流研究成果来看，最常见的行为识别系统结构包括几个模块：目标检测、目标跟踪、行为识别。本文首先对图像进行判断是否有人体目标，识别出人体后对图像进行灰度化，在对灰度图像用背景差法与背景图像比对，最后，比对提取出的人体来判断人体处于何种行为。

关键词：matlab，肤色识别，行为检测

Human behavior detection and recognition

Abstract

Matlab human behavior recognition and detection of computer vision, intelligent video surveillance, human motion analysis, the nature of the interaction, virtual application prospect and reality of the economic value of the field, so a lot of research a hot topic these problem areas. Because of the complexity and diversity of human movement of the external environment, so that human behavior recognition has some challenges. The study of human behavior in its infancy now, pending further study and discussion.

Recognition matlab studied human behavior recognition and detection based on paper studies is judged from the image in which the target behavior, such as a common walk, sit, lie down, running, jumping and other acts. From the many existing mainstream research point of view, the most common gesture recognition system architecture consists of several modules: target detection, target tracking, behavior recognition. Firstly, the images to determine whether there are human targets identified after the body of the gray-scale image, the gray-scale images using background subtraction and background image comparison, and finally, more than the extracted human body is what determines kind gesture.

Keywords: matlab, color identification, behavior detection