3D Scanner using Kinect

In this report, we describe, justify and assess the quality of our design for a 3D scanner using Microsoft Kinect. A 3D model of an object can be obtained by taking both color and depth images from different viewpoints. We developed a streamline and a novel combination algorithm for this task. In the process, we had to solve problems involving noise, miss-alignment between color and depth, combination of different viewpoints and meshing. The final mesh can be converted to a format that can be fed into the RepRap 3D printer. We provide justifications for some of our major design choices such as meshing methods, combination, alignment. And we also assess the quality of our final design with respect to the revised requirements and find that the final design satisfies all our original requirements. Appendix A provides details of our combination algorithm, appendix chap:meshsoft provides an overview of software that take pointclouds to meshes, and appendix C provides an overview on the alignment procedure dealing with color/depth mismatch. We thank Orbis Mutual Funds for sponsoring this course and Prof. Jason Foster for his enthusiasm and helpful suggestions throughout this course.