Multiview registration-based handheld 3D profiling system using visual navigation and structured light

Abstract

This article describes the 3D handheld profiling system composed of a stereo camera and an illumination projector to collect high-resolution data for close range of applications. Visual navigation approach is either based on feature matching or on accurate target, and the target-based approach was found to be more accurate if the 3D object has less texture on its surface. Block matching algorithm was used to render the single-view 3D reconstruction. For multiview 3D modeling, coarse registration and final refinement of the point clouds using iterative closest point algorithm were utilized. The proposed approach yields good accuracy for multiview registration as demonstrated in the results of this research.