Reconstruction of typhoon path and cloud image from descriptors

In this paper, we propose a novel method to reconstruct the path and contour of a typhoon in a satellite image using mathematical morphology and descriptors. Generally, it is hard to depict the contour and path of a typhoon by a simple way in the satellite images. Especially, if we would like to track changes of a typhoon in the visual successive movement form on a browser or mobile device through the network in real time, it is necessary to reduce the resolution of satellite images. Our proposed method makes it possible to carry out such a task without complicated computing. First, we apply features extracted from the infrared (IR) satellite cloud images to identify and locate typhoons automatically. Next, we reconstruct an approximate contour and reliable movement path of a typhoon by cloud pattern descriptors. Our proposed method simplifies the feature representation of a typhoon shape and improves the efficiency of the typhoon contour reconstruction. Moreover, the proposed method can be applied to improve analysis and prediction of typhoon trace. We used our proposed method to track typhoon paths and reconstruct typhoon contour images occurred in recent years and achieved some satisfactory results.