Determining the depth measurement for the 3D model reconstruction

Abstract

Abstract: This paper deals with an efficient method for estimating the depth measurement of an object from the 2D images. The quadranocular stereo system has been developed based on the multiple baseline stereo approach with the perpendicular camera positions. The disparities among the multiple stereo pairs are computed from the corresponding pixel pair of an object. The dense depth map is generated from the disparities of the stereo images grabbed by this camera system. The 3D data positions of the object are computed from the depth map. The 3D virtual views of an object are synthesized using its depth data and color information. The experiment with real images has been performed to confirm theeffectiveness of the proposed method. Keywords : 3D virtual view, synthe sized image, baseline stereo, depth measurement, disparity image, occlusion errors 1. Introduction The 3D modeling technology occupies theessential role in design and manufacturing fields as shape modeling, computer graphic (CG), CAD, CAM and CAE systems. The precise depthmeasurement isimportant in the 3D modelreconstruction work. Furthermore,the accuratedepth measurement provides the constructionquality of the synthesized 3D view in virtualized reality. Reconstructions from the single stereo pairoften have occlusion errors due to the cameraalignment. For reducing the occlusion errors and ambiguity in matching, the multiple stereo pairs are applied in many recent research works