AETC: Segmentation and classification of the oil spills from SAR imagery

Identification of oil spills from synthetic aperture radar (SAR) images is a complex task that has recently come into use. Existing oil spill detection methods have been found to be expensive and complex and require high processing power and time. Also, noise removal and extracting the features of oil-spill images are major issues. To overcome these drawbacks, a novel segmentation technique, adaptive edge and texture clustering (AETC), is proposed in this article to detect and classify the oil-spill area in a given image. The proposed technique contains four main stages: preprocessing, segmentation, feature extraction, and classification. Initially, the input image is preprocessed to eliminate speckle noise and to enhance the quality of the image by using the Gaussian distribution function. After that, the preprocessed image is segmented with the help of the linear edge weighted (LEW) clustering technique. In this stage, the boundary region is identified and the contour is segmented. The features of the segmented image are extracted using the convoluted horizontal vertical (CHV) pattern extraction technique. Finally, the relevance vector machine (RVM) classification technique is applied to classify the oil spill portions from the given image. The experimental results evaluate the performance of the proposed system in terms of accuracy, sensitivity, specificity, Jaccard, Dice, and Hausdorff distance. Here, the existing back scatter-gradient-artificial neural network technique is compared with the proposed CHV-RVM technique. From this analysis, it is proved that the proposed techniques provide the best results.