Satellite Image Segmentation and Classification Using Fuzzy C-Means Clustering and Support Vector Machine Classifier

Abstract

Feature extraction and classification are important areas of research in image processing and computer vision with an extreme great number of applications in science and industry. One of the applications is satellite image classification. The main objective of this research work is to study image segmentation, feature extraction, and image classification algorithm, apply it on satellite images to classify residential, mountain, forest, desert, and river region. Automated segmentation is performed using fuzzy C-means technique and study and identify technique for classification of regions in satellite images. Features are extracted from the segmented output image using grey level co-occurrence matrix (GLCM) and Gabor filter. Then, the classification is used to classify the region using SVM classifier based on feature extracted. We can propose work based on colour and texture features with the help of Gabor filter and GLCM which are more appropriate in extracting colour and texture features. As a result, the number of features will increase accordingly classification accuracy will also get increased. We can also add one more phase which is image segmentation for more accurate results. We can segment the input image using fuzzy C-means clustering technique and after that we can perform the feature extraction on the segmented image, in this way, we will get the results which are more optimized and takes less time to extract feature than the previous work. This work can be enhanced in context of classification accuracy and enhance feature extraction phase.