Surface Extraction from Multiple Views with Different Zoom Ratio

We propose a high-resolution digitizing approach to the extraction of shape and surface reflectance of a real 3D object. The object is set on a rotary table and taken by a HDTV camera. The camera and the rotation axis of rotary table are well calibrated. The 3D object's shape is recovered through stereoscopic measurement using multiple images by rotating the rotary table horizontally and the camera vertically. To measure a region with a narrow section, we employed the images taken with a long focal length to estimate its shape. The entire 3D shape is obtained by merging all point data in 3D space through the estimated zoom ratio. A surface triangle mesh is generated to represent the object shape. For each 3D object, the intensity variation data from all views is computed for estimating surface reflectance. Experimental results are presented to demonstrate the effectiveness of our approach.