期刊名称：International Journal of Computer Science & Technology
印刷版ISSN：2229-4333
电子版ISSN：0976-8491
出版年度：2012
卷号：3
期号：1Ver 3
出版社：Ayushmaan Technologies
摘要：This paper describes the object/region based approach for satellite image classification to extract snow cover area in high mountainous region. The objective is to develop an application to classify the satellite imagery using a Fuzzy-based supervised classification method using spectral information of an input satellite image. The procedure first divided the image into several groups (in the spectral space) based on selected training datasets using the Gaussian function as a measure of similarity. The result was then segmented into regions using fuzzyfication process. The fuzzyfication process also incorporates edge information to avoid intermixing of sub-pixels of remotely sensed images. After that the extracted regions has been classified using Gaussian function rule. Several regions were selected as training samples for region classification. Each region has been compared to the training samples and was assigned to its closest class. The procedure has been implemented using MATLAB software and tested on USA’s MODIS TERRA imagery. This procedure significantly reduces the mixed pixel problem which was suffered by most pixel based classification methods.
