Publication

Abstract : Texture investigation is a broad field of study with applications extending from remote sensing, satellite communication and autonomous systems to advanced systems such as robotics and machine learning. Textural images define the pattern of pixels spatially arranged, where a specific order can be used to identify and classify images. Human level analysis and classification of textural images have been challenged by current technical advancements with similar level of accuracy. Machines learn better when the images are more distorted and can even spot minute differences in parameters. Efficient classification of images is ensured if the parameter governing the decision is robust and exclusive. Statistical features extracted from textural images are learned through support vector machines, and the learned database is used with testing images to obtain the accuracy of image classification.