基于图像的人体参数测量系统的设计与实现  

Design and Implementation of Human Size Measurement System Based on Images

作　　者：吴壮志[1] 廖爽爽[1] 聂磊[1] 王政[2] 王春慧[2] 周诗华[2] 

机构地区：[1]北京航空航天大学计算机学院,北京100191 [2]中国航天员科研训练中心,北京100094

出　　处：《湖南大学学报（自然科学版）》2010年第9期88-92,共5页Journal of Hunan University:Natural Sciences

基　　金：国家科技支撑计划资助项目(2008BAH29B02)

摘　　要：给出一种基于图像的人体参数自动测量方法,并设计与实现人体参数测量系统.通过2个数码相机拍摄人体的正视图和侧视图,运用图像分割和特征提取等图像处理方法提取2幅图像上的人体特征点;利用三维标定架标定相机的内外参数;利用双目视觉原理从2幅图像上的特征点计算出人体的测量点,完成人体尺寸参数计算.试验结果表明:与传统手工测量相比,本方法能实现自动测量,缩短了测量时间,提高测量的精度和效率;与三维非接触测量相比,本方法测量设备简单和便携,测量速度更快,操作更加方便.This paper presented a human size measurement method based on images, and designed and implemented a measurement system. Firstly, the front and side images of a tester were captured by using two digital cameras simultaneously. Secondly, the body feature points of two images were extracted in the image segmentation and feature extraction methods. Thirdly, the intrinsic and extrinsic matrixes of two cameras were calibrated. In the end, the body measure points were calculated through the feature points of two images based on the binocular vision principle, and then the body measure items were calculated. The experiments have shown that, compared withe the traditional size measurement methods, this method is characterized by automatic measurement, short measurement time and flexible operation. Compared with the 3D non-contact measurement techniques, this method features cheap and flexible equipment, and rapid speed and concise working process.

关 键 词：人体测量学 双目视觉 相机标定 特征点匹配 图像分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]