3D Object pose estimation from binarized images

This paper discusses a 3D object pose estimation problem in which binarized images are used. Assuming that a CAD model of the object of interest is available, minimization of the difference between the captured and rendered images is considered. To make this minimization feasible in the situation where only binarized images are available, we propose to render a higher-resolution image at first and then resize it to the original size with help of an anti-alias filter. The simulated results show that the proposed method performs well as long as the given image offers sufficient clues.