Collective Calibration of Active Camera Groups

Until recently, traditional approaches to the task ofcamera calibration have relied on the use of accurategrid patterns, or strategically placed targets. Suchapproaches can prove time consuming and oftenrequire expert supervision. For ambient intelligencetype applications, it is unwise to rely on theconsistency of camera positions and orientations;consequently a fully autonomous camera calibrationprocedure is preferred. In this paper we present an auto-calibration systemfor the estimation of the extrinsic properties of activecameras lying within indoor, self-observable groups.Neither prior knowledge of camera locations, nor acalibration-pattern is required, just a few basicparameters concerning the physical appearance of thecameras. Groups of active cameras can be calibratedwithin minutes, by exploiting their precise controlmechanisms. As no supervision is required, cameradeployment configurations can be changed or newcameras added easily.