家用陪护机器人人脸检测与识别系统研究

发布时间：2018-03-01 22:34

本文关键词： 家用陪护机器人人脸检测人脸识别 Adaboost Retinex　出处：《昆明理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着经济和科技的发展,机器人技术及其相关产业也迅速拉开了发展的序幕,机器人种类和技术的不断创新,使得机器人逐步扩展应用于各个领域。作为机器人的一个种类——服务型机器人,以为人类提供个性化和人性化服务为宗旨,目前在研究者们不断努力创新下,服务型机器人己经向越来越智能化的方向前进。家用陪护机器人人脸检测与识别系统作为家用陪护机器人视觉系统的重要组成部分,是实现机器人与人友好交互的一种关键技术。本文研究了家用陪护机器人人脸检测与识别问题,主要做了如下工作:以Haar-like特征遍历图像提取特征,采用高效快速的Adaboost算法对特征筛选,实现该算法训练的级联分类器对摄像头采集的图像检测人脸。利用Gabor小波提取人脸特征和PCA低维空间仍可以有效表达人脸特征的二者优势相结合,有效提高了人脸识别率。针对光照遮挡人脸对检测与识别性能的影响,通过对比Retinex相关算法对图像光照预处理的效果比较,采用了改进的Retinex算法对采集的图像进行光照预处理,通过实验结果表明有效地提高了光照遮挡人脸情况下的人脸检测率,并在一定程度上提高了人脸识别率。在复杂光照情况下人脸检测与识别技术的理论研究和实践的基础上设计了家用陪护机器视觉整体架构,构建了软件开发平台,本文研究的硬件平台是上海合时公司的家用陪护机器人,把人脸检测与识别技术应用于移植在移动控制端,实现了家用陪护机器人的快速人脸检测与识别,完成了系统软件各功能模块的设计、编写和测试,在机器人更加智能化方面做了尝试。
[Abstract]:With the development of economy and science and technology, the robot technology and its related industries have also opened the prelude of the development rapidly, the kind of robot and the continuous innovation of technology, As a kind of robot, service-oriented robot, as a kind of robot, which aims to provide personalized and human service for human beings, has been continuously innovated by researchers. Service-oriented robot has been moving towards more and more intelligent direction. The face detection and recognition system of home escort robot is an important part of the vision system of household escort robot. It is a key technology to realize the friendly interaction between robots and people. In this paper, the problem of face detection and recognition of home escort robot is studied. The main work is as follows: traversing the image with Haar-like features to extract features, Using efficient and fast Adaboost algorithm to filter features, The concatenated classifier trained by the algorithm detects the face of the images collected by the camera, and combines the advantages of extracting face features with Gabor wavelet and PCA low dimensional space to express face features effectively. Aiming at the influence of illumination occlusion on the detection and recognition performance, the improved Retinex algorithm is used to preprocess the collected image by comparing the effect of Retinex correlation algorithm on image illumination preprocessing. The experimental results show that the face detection rate under illumination occlusion is improved effectively. On the basis of the theoretical research and practice of face detection and recognition technology under complex illumination, the overall architecture of home escort machine vision is designed, and the software development platform is constructed. The hardware platform studied in this paper is the home escort robot of Shanghai Heshi Company. The face detection and recognition technology is applied to the mobile control terminal to realize the fast face detection and recognition of the home escort robot. The design, programming and testing of each functional module of the system software are completed, and a more intelligent robot is tried.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【参考文献】