Contented Based Cloudy Satellite Images Retrieval System Design And Practice

Satellite Cloudy Images are very useful in weather forecasting. At present, cloudy images retrieval systems almost based on text. However it is very difficulty to describe an image in word, so this kind of retrieval systems is not very efficient.This dissertation makes a systematic and detailed study on Contented Based Cloudy Satellite Images Retrieval System and builds a prototype system to test the theory. After carefully study of images received from weather satellites, it is obviously that color, texture ,shape and spatial features will help us to distinguish different images. This paper use color and texture features, which was extracted from satellite images, to represent unique images. By using Gray Histogram and Co-occurrence Matrix, images can have their unique feature vectors. Then restore these vectors in satellite images database. After Normalization, all the similarity distance values are within the range of [0,1]. With these unique features, user can retrieval the most similarity images from images database which has already been classified. The prototype is based on Jbuilder X and MySQL 4.0.20d.The main achievement in this as below: This paper adopt Contented-Based Image Retrieval method in cloud retrieval. Then design and accomplish a retrieval system prototype. This efficient prototype has a better interface between human and machine. After analyzing of satellite cloudy images, this paper choose color feature and texture feature to build unique feature for represent image. In the end, the prototype system computes image's multifeature value by using different weight.