F.Image Feature DetectionFeature re

F.Image Feature DetectionFeature recognition is a topic of increasing interest with the recent developments in deep learning. The ability to distinguish various features in the image, such as buildings, roads, vegetation and other categories, could be of help in various applications such as improving urban planning, environment monitoring, identifying the disaster prone areas and so on. Edge detection is an important application in the field of image processing. Currently various edge detection techniques such as PSO , Preweitt, Laplacian and Laplacian of Gaussian are being used. The disadvantage of this technique is that the edge thickness is fixed and implementation of threshold is difficult. An efficient oil spill detection in sea is required by the International Maritime Organization to carry out evaluation of trade routes through sea. This requires a feature selection based on machine learning algorithm [16]. The SAR images are widely used to provide images on the oil spills on the surface of the ocean. It can provide wide area coverage and good image clarity. The oil spill detection has three main stages. In the first stage, the feature is separated from the background using dark spot segmentation. In the second stage, feature extraction is done to get the feature vectors which contain information to distinguish oil spills from the rest of the image. In the third stage, dark spot classification is done where the distinction criteria between the feature and non-feature regions is performed. Feature Selection is the most prominent step in the machine learning process which creates a set of features which clearly describe the detection problem. One of the main drawbacks faced is the presence of the clouds which affect the image quality in satellite images. This makes it difficult for the researcher who is trying to get some information regarding the terrain. Hence cloud detection becomes necessary as part of satellite image processing. The panchromatic satellite images make the cloud detection difficult since the cloud distribution is highly irregular.