Certain investigations on feature extraction and indexing techniques in content based image retrieval

Abstract

Internet traffic refers to the flow of data within the entire internet and file sharing constitutes a large fraction of internet traffic. This internet traffic is mostly dominated by the multimedia content and it is predicted to intensify more in the future. This remarkable growth of multimedia data makes it challenging to store and retrieve them efficiently. Thanks to Moore s law, the electronic gadgets and devices have become smaller and smarter in terms of storage but when it comes to effective retrieval, they are still years behind. Content Based Image Retrieval (CBIR) aims at retrieving digital images from large image databases using computer vision techniques. Content Based Image Retrieval finds applications in various fields including art collections, intellectual property maintenance, crime prevention, remote sensing systems, geographical information handling, military, medical diagnosis, architectural and engineering proposals etc. The purpose of an image database is to store and retrieve an image that is pertinent to the user query. Some image collections either in local storage or in the internet are highly structured in terms of keywords, thereby making text-based search effectual for finding a specific image, or images with a specific content. Nevertheless, most of the digital images are poorly labelled or not labelled at all. With the volume of images increasing, it is essential to find other ways of searching images. In contrast to text-based search, this study must aim to think of tools that can directly look into images and retrieve them based on the content instead of keywords. The method of organizing and retrieving the images based on the content is referred to as Content Based Image Retrieval (CBIR). newline