New Rectification Technique for Cloud Base Height Estimation Base on Stereo Cameras

Abstract

A Whole Sky Imager (WSI) has a single camera to take photos of the cloud from the ground. It can provide only the cloud cover index and cloud direction. It cannot give out the cloud base height or speed. In this paper, a pair of WSI is synchronized to capture stereo images. We propose a new rectification technique that is a part of WSI-installation which directly effects Cloud Base Height estimation. the semi-global matching technique is proposed to find matching points on both images. Triangulation can then yield the distance of the cloud base. Calibration and testing of the proposed system are done by comparing the distances of known buildings estimated by our analysis with the distances obtained by Google Maps. The average errors are about 4% for targets in the range of 150m, and 15% for targets in the range of 1km.