Meteorological image processing with wavelets

The meteorological image system deals with images taken by a satellite in the following three channels: the visible, the infra-red and the water-vapor channels. This imagery is relevant for many applications, particularly weather prediction. To make a good weather prediction, we need a quite high amount of images frames to cover the number of observations required. It is, therefore, extremely important that the image acquisition system incorporates adequate capabilities that enable it to compress this huge amount of available data in an efficient way. A second problem usually come across when dealing with meteorological images is to find the best way to identify the important information that has to be extracted for recognition purposes. This is commonly done by an edge detection or texture discrimination process. A third problem concerns noise removal. The resolution of all theses problems is closely related to the way of representing the signal. In this respect, recent research work shows that a signal representation by way of the wavelet transform seems to have some interesting properties and promises to yield high compression ratios. The present paper is an attempt to attain conclusive results.