렌덤 패턴 투영을 이용한 스테레오 비전 시스템 기반 3차원 기하모델 생성

Abstract

3D geometric modeling of an object of interest has been intensively investigated in many fields including CAD/CAM and computer graphics. Traditionally, CAD and geometric modeling tools are widely used to create geometric models that have nearly the same shape of 3D real objects or satisfy designers’ intent. Recently, with the help of the reverse engineering (RE) technology, we can acquire 3D point data from the objects and create 3D geometric models that perfectly fit the scanned data more easily and fast.

In this paper, we present an approach for the generation of 3D geometric models based on a stereo vision system (SVS) using random pattern projection. In order to obtain reasonable results with the SVS-based geometric model generation, we deal with many steps including stereo vision system configuration, camera calibration, stereo matching, scanning from multiple views, noise handling, registration, merging, and triangular mesh generation. A triangular mesh is considered as the resulting geometric model.

We first build up the measurement system, which is low-priced and easy to use. The system is composed of two cameras, two frame grabbers and a beam projector. We properly integrate stereo vision and shape construction techniques for smooth transition of data from measurement to registration and merging. We present a method for acquiring hight-quality stereo image matching with pixel-accurate correspondence information using random patterns projection. In order to acquire more accurate stereo matching, we project random patterns onto the object's surface using an LCD projector. We also describe efficient algorithms for processing measured 3D data including noise handling, registration and merging.

With experiments using various random patterns, we have found that the proposed approach can generate 3D geometric models in good quality. Some examples are given to show the quality and usefulness of the approach.