Study of CBIR methods for retrieval of digital images based on colour and texture extraction

An image retrieval system is a software system which is used to browse, search and retrieve images from a large database of digital images. It is a specialized search to find digital images. In most applications of image processing, it becomes necessary to find images by using text, keywords or by using any other features of an image. The conventional or traditional methods of image retrieval includes addition of metadata like captions, important keywords to describe image etc by using which retrieval of images takes place. The manual annotation of images is highly time-consuming, laborious and costly in terms of investment of money. Thus, automatic image annotation is important and there are various techniques to do so. The increasing web applications and the semantic web have encouraged the development of tools for automatic image annotation. When an image is searched on the web, it is really a tedious task to retrieve any image as the database is huge. Therefore, Content-based Image Retrieval (CBIR) i.e. Query by Image Content (QIC) or Content-based Visual Information Retrieval (CBVIR) is used. It is a computer vision based approach to address the image retrieval problem. The content of an image (colour, text, shape or any other information etc) is used to retrieve any information in the visual form. This paper presents a theoretical study and practical implementation of CBIR methods based on colour and texture carried out in MATLAB, and the results of the implementation are stated and their comparison is made. The objective of the work reported in this paper is to practically study CBIR methods and understand how the visual information can be retrieved using them.