SATELLITE IMAGE ANALYSIS USING CONTEXTUAL DATA RETRIEVAL TECHNIQUE FOR ENVIRONMENTAL MONITORING CONDITION

摘要

Satellite images show specific regions depending on various angle or locations based landscape image which contains different colour regions from a top angle view. Many image processing approaches use the transform mathematical function to classify satellite images nowadays. In the previous method, Convolutional neural network (CNN) approach has been used to classify the satellite image. However, those methods have several drawbacks such as output inaccuracy and difficulties in identifying the extracted land area of image for remote sensing application. Hence, Contextual data retrieval technique (CDRT) with different landscape images for environment application. This method consists of four stages: input image, preprocessing, segmentation, and classification. In this system, input of satellite images analysis is taken and from that analysis, the specific input image region is enhanced and the input image is preprocessed using median filter for noise free image. The segmentation using Multi divergent clustering technique (MDCT) separates different regions like land, water, forest area, from that findings, the agriculture land, water land area and different layers are classified based on image analysis taken for environment application. Classification using CDRT methodology is done to find the extracted layers and to give better accuracy with MATLAB17 tool. The classification result of the proposed method is 94.55% when compared to existing methods as the proposed method gives better output result.