Image-based 3D Model Construction

• Main Authors: Kuan-Chen Chen, 陳冠臻
• Other Authors: John Y. Chiang
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/00914297739608905840

碩士 === 國立中山大學 === 資訊工程學系研究所 === 89 === The shape construction of three-dimensional objects has numerous applications in area that include manufacturing, virtual simulation, science, medicine, and consumer marketing. In this thesis, we consider a automatic system which captures and triangulates views of a real world 3D objects and finally registers and integrates them. There are four steps in our system, image acquisition, image processing, model construction and stereoscopic display. First step, image acquisition, we take 2D image pairs (by CCD camera movement) from different angles of the model with one CCD camera. Second step, image processing. In order to derive depth form two images captured by the CCD camera, we find registration points between two images by using image segmentation, feature extraction, image registration. Third step, 3D model construction, we divide it into three parts. First part, we generate partial depth surface by Delaunay triangle splitting , and for a selected set of viewing directions. Second part of this step, different surfaces have to be mapped into a uniform coordinate system for the given 3D object. Integration of registered surfaces defines the third part of the model construction. This finally, can lead to the generation of a complete 3D model of the given scene or of the given object. After generating a complete 3D model, we create a stereoscopic view in last step. We put on the LC-Shutter-Glasses and look through the lenses at high-resolution full color display while the lenses "shutter" on and off alternatively. The monitor displays only the left view while the right lens of the glasses shutters, and display the right view while the left lens of the glasses shutters.