This paper introduces the design and implementation of a foot measuring system based on stereo vision. This system uses several cameras to capture images, then transforms information of 2D images into 3D data. This system consists of three parts: image capturing, camera calibration and modeling. In this paper, we focus our research emphasis on several unavoidable issues of stereo-vision based measuring, such as camera calibration, triangulation in modeling.This system implements a hardware platform of foot measuring, which configures cameras based foot's feature and collects foot images. Image capturing module could dynamically control the hardware platform and real-time capture images;camera calibration module measures the intrinsic and extrinsic parameters of camera;modeling module deals with the images which are captured by hardware platform, detects corners, matches these points, and finally reconstructs three-dimension foot model.This paper introduces the work process of whole system, the theories of system implementation, include camera calibration, corner detection, feature points matching, stereo vision based 3D reconstruction and triangulation. The paper explains the camera calibration method used in the system and the calibration method under wide angle lenses, and an automatic triangular meshes sewing algorithm. In addition, the paper also shows some experiment results of calibration and triangular meshes sewing. At last, the paper summarizes the work has been done, and expects the future developing. |