Outdoor camera calibration method for a GPS & camera based surveillance system

A reliable outdoor binocular camera calibration method for a multi-GPS (Global Positioning System) apparatuses and multi-cameras based surveillance system is introduced in this paper. Different to other system, the user needs our system to track the motion and estimate the shape of the selected vehicle in real time by our binocular cameras. So we implement our system by these steps below: 1) fixing a GPS apparatus in each vehicle to collect the location information. 2) Calibrating the interior and exterior parameters of the binocular cameras. 3) Computing the coordinate transform matrix between the GPS coordinate system and the camera coordinate system. 4) Using the GPS information to guide the binocular cameras to track the moving vehicle. To calibrate the binocular camera in an outdoor environment, we use the Scale Invariant Feature Transform (SIFT) descriptor to compute the features. The RANSAC matching technique is used to eliminate the mismatch point pairs. After that, we use an essential matrix based self-calibration technique to compute the camera parameter. Finally, we analyze the relationship between the blur affection level of the calibration image and our proposed method. Many experiment results have shown the robustness of our method.