期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2008
卷号：XXXVII Part B5
页码：57-62
出版社：Copernicus Publications
摘要：An automatic 3D measurement and modelling method based on stereo-camera was proposed in this paper. Firstly a hardware system for data capturing, named stereo-camera system, was developed. This system included three cameras, which were fixed by joint poles between them. Cameras' intrinsic parameters and relatively extrinsic parameters between them could be obtained by calibration. With the help of stereo-cameras three images of the object can be captured at the same time and then processed with digital photogrammetric method to produce local point cloud. Meanwhile texture data could be picked up from the middle image and the textured local 3D surface model can be constructed. All local models were produced by the same process. Then a whole adjustment optimization method was developed to register all the local models into a uniform coordinate system with the corresponding point information between adjacent local models, and finally the entire 3D model could be constructed. This method has the same function as laser scanner. However, compared with laser scanner, this method is low-cost. In addition, texture can be extracted from original images and textured model can be reconstructed easily because it is image-based method. Instead of two cameras, this system used three cameras, so high reliability can be achieved. Its main shortcoming is that sometimes it fails for objects lacking texture. Experiment results prove above conclusions
