Categorization of Images Using Content-Based Features: A Data Mining Approach.

Date of Submission

December 2007

Date of Award

Winter 12-12-2008

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Electronics and Communication Sciences Unit (ECSU-Kolkata)


Bagchi, Aditya (CSSC-Kolkata; ISI)

Abstract (Summary of the Work)

Images are being extensively used in every sphere of our life. Apart from overwhelming influence of television, common people look for images in newspapers, advertisements, item catalogues, entertainment, education, architecture, painting and many others. Professionals use image in criminology (e.g., fingerprint identification, face recognition), medicine (e.g., case-based diagnosis from radiographs or scan data), education (e.g., searching for material in Library), fashion design, historical archiving, fine arts and so on. Most of the cases the problem is to find a desired image from a large collection or, in other words, retrieve images similar to the image at hand from large number available in some collections. Image search and retrieval is a field of very active research since the 1970’s. However, the field has observed a steady exponential growth in recent years as a result of unparalleled increase in the volume of digital images. Thousands of images are generated everyday for different applications. These images are either stored in a local database or are available from remote ones. Thus a huge amount of information is out there and can easily be accessed through world-wide web. Professionals of various fields intend to access and utilize these images for their purpose. However, we cannot access to or make use of the information unless it is properly organized for efficient browsing and retrieval, because searching and locating a desired piece of image from varied and large collection usually result in a total frustration. Two major research communities, namely Database Management and Computer Vision, are putting considerable effort towards the solution of this problem. Accordingly two major approaches have emerged: one being text based and the other visual based respectively.Early systems of image retrieval exploited the capabilities of text based Database management Systems. Images are first manually annotated using a set of keywords that describe the content of the image best. Images are indexed and arranged using these keywords, finally images are retrieved based on text based query. Major research in this direction includes Data Modeling, Indexing Structure, Multi-dimensional Indexing, Efficient Searching and Query Design and Evaluation. However, these text-based image retrieval techniques face two major problems: labor intensiveness and annotation impreciseness. When image collection is large, enormous amount of man-hour is required to annotate those images manually. Problem became more and more acute since early 1990’s when world-wide web allow access remotely placed image databases. The second problem is more crucial and is due to semantic of image content. Because of rich content in the images and the subjectivity of human perception, same image may be perceived differently by different persons. As a result, same image may be annotated by different set of keywords by different persons. Thus image annotation in general is neither unique nor adequate; hence affects the performance of image retrieval system to a large extent. This leads to development and flourishing the alternate approach, namely Content Based Image Retrieval (CBIR) system1.2 What is CBIR? Content-based image retrieval (CBIR), also known as query by image content (QBIC) and content-based visual information retrieval (CBVIR) is the application of computer vision to the image retrieval problem, that is, the problem of searching for digital images in large databases. The term CBIR seems to have originated in 1992, when it was used by T. Kato to describe experiments into automatic retrieval of images from a database, based on the colors and shapes present. Since then, the term has been used to describe the process of retrieving desired images from a large collection on the basis of syntactical image features. The techniques, tools and algorithms that are used originate from fields such as statistics, pattern recognition, signal processing, and computer vision.In CBIR systems the term “content-based” means that the search will analyze the actual contents of the image. The term 'content' in this context might refer colors, shapes, textures, or any other information that can be derived from the image itself. Without the ability to examine image content, searches must rely on metadata such as captions or keywords, which may be laborious or expensive to produce. There is growing interest in CBIR because of the limitations inherent in metadata-based systems, as well as the large range of possible uses for efficient image retrieval. Textual information about images can be easily searched using existing technology, but requires humans to personally describe every image in the database. This is impractical for very large databases, or for images that are generated automatically, e.g. from surveillance cameras. It is also possible to miss images that use different synonyms in their descriptions. Systems based on categorizing images in semantic classes like "cat" as a subclass of "animal" avoid this problem but still face the same scaling issues.


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


This document is currently not available here.