An Efficient Method for the Retrieval of Content Based Remote Sensing Images

Abstract

Image retrieval is an important area of research in the field of image processing. This process enables digital image collections to be generated quickly and made accessible across the World Wide Web (WWW) to multitudes of users. The images are retrieved by their properties such as color, texture, and shape etc. Content-Based Image Retrieval (CBIR) is nothing but retrieval of a large number of images based on the content of the image. It is more advantageous than the Text Based Image Retrieval (TBIR). Content is nothing but features of the image. For Large databases, the images are retrieved using CBIR. It plays an important role in remote sensing applications. Sensory distance is the difference between the objects in the environment, which is calculated from details and extracted in a defined format. The semantic gap is the lack of consistency between information which is derived from visual data and the understanding of the same data in a given situation by the user. Remote sensing images are images taken from satellites that are able to do what ten thousand words normally do. Satellite images do immense service to every nation. They find great use in the field of agriculture, urban planning, and meteorology. The problem is to develop a CBIR system, which learns about the existing semantic categories in the training dataset, using the convolutional neural learning concept, Support Vector Machine (SVM). In remote sensing image retrieval, the color feature is used for extracting a feature from the image. The color feature is also called as the visual feature. In the probability distribution of colors, the statistical moments shown are the color moments, which are used in the retrieval systems. The parameters, which are calculated in this method are Variance, Mean, and Skewness. newline newlineFor the trained category, and the untrained category of images the system must provide the correct retrieval results. Hence, the problem is extended to find an adaptive learning scheme which would adaptively learn about the new category