Calibration of Surveillance Cameras based on an Auxiliary Color-Depth Camera

Camera networks provide large-scale video surveillance due to the summation of the fields-of-view. Usually, the fieldsof-view are non-overlapping making difficult the tracking of targets even when their motion speed would allow predicting their future locations. That requires a global calibration methodology, encompassing both the intrinsic and extrinsic parameters of each camera on a common reference frame. This thesis is focused on the specific aspect of using an auxiliary color-depth camera to calibrate one camera at a time. The data acquisition device to perform this global calibration, consists of a pre-calibrated Pan-Tilt-Zoom (PTZ) camera with a laser pointer. This setup allows acquiring point clouds, on the areas corresponding to fields-of-view of the fixed cameras. The calibration methods are based on detecting and corresponding SIFT features and estimating the calibration parameters based on optimization methodologies, namely DLT-Lines and DLTPoints.