The Transformation of Range Data Based on the Combined Application of Image Segmentation and NURBS Surface

概要

論文の詳細を見る
A real object is often described by its geometric surface and color image. The surface properties and color features sometimes correspond with each other. This paper introduces a new method for the redesigning of 3D model based on the combinedapplication of image processing and computer graphics. In particular, we focus on the 2D planar images and 3D range data of a real object, which are produced by a laser range finder and determined in an identical coordinates system. Such a method allows us to create a variety of complex models from an original model. Image processing is used to specify the features of a real object according to its RGB values so as to avoid the inconvenience of analyzing surface properties. We first detect the feature region through image processing, for example, color thresholding, edge detection, and labeling. Then, we displace the original range data within the detected region with nonuniform rational B-spline (NURBS) surface data. Finally, we reconstruct the new range data set wlth a smooth subdivision surface and map it with an original image. From just one real object, this method can provide a number of transformed models.
日本バーチャルリアリティ学会,東京大学大学院工学系研究科の論文