Automatic sphere detection for extrinsic calibration of multiple RGBD cameras

RGBD cameras provide color and depth images at a high speed, so they are becoming increasingly important in many computer vision applications. With the color and depth information, it is easy to build a textured 3D model of a scene without effort on correspondence search. To combine depth data from multiple RGBD cameras in a unified coordinate system, the cameras need to be extrinsically calibrated based on 3D correspondences of scene features. To maximize the scene features simultaneously observed from different viewpoints, a spherical calibration object is widely used. In this paper, we propose a method for automatically detecting a spherical object in an RGB image. Since the detection relies only on RGB information, the proposed method has a potential to be extended to conventional multi-view calibration. Experimental results show that the proposed method accurately detects spherical objects in a cluttered environment.