基于多目视觉的变形钢板三维测量方法  

3D Measurement Method of Deformable Steel Surface Based on Multi-Vision

作　　者：史金龙[1] 白素琴[1] 谈友霞[1] 丁建雯[1] 刘庆华[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《计算机测量与控制》2011年第9期2119-2121,共3页Computer Measurement &Control

基　　金：国家自然科学基金(51008143);江苏省汽车工程重点实验室开放基金项目(QC201005)

摘　　要：提出了一种使用4个相机测量变形钢板三维形状的方法;在没有纹理的钢板上画一些点,使用4个相机同时拍摄钢板;利用边缘检测算法找出钢板的边缘,利用特征点提取算法提取出4个图像中钢板上的特征点,并找到4个图像上精确的特征点匹配关系,然后测量出这些特征点的三维点;最后,以这些三维点为种子点向外区域增长,从而得到钢板的三维形状;实验对变形钢板进行三维测量,得到了钢板的精确的三维形状,结果验证了该方法的有效性和可行性。This paper presents a method for measurement of deformable steel plate using four cameras. If the plate is texture less, we need draw some points on the steel plate. First, four cameras are used to capture the images of steel plate; Second, edge detection algorithm is adopted to find the edge of steel plate in images; Third, feature points on images are extracted by feature extraction algorithm and the cor- respondences of feature points among the fours images are found. According to the matching of feature points, 3D points are measured. Fi- nally, region growing is implemented based on the measured 3D points, and 3D shape of steel plate is obtained. The performance and effec- tiveness are evaluated by simulated and real data experiments.

关 键 词：视觉 三维测量 变形钢板 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]