Conversion of an image to a document using grid-based decomposition for efficient content based image retrieval
Dinesha Chathurani Nanayakkara Wasam Uluwitige, Shlomo Geva, Vinod Chandran · QUT ePrints (Queensland University of Technology) · 2015
Initially this work focuses on identifying various sizes of semantic image feature building blocks which can be used to represent an image as a bag of semantic image features. Then a modified Bag-of-Visual Words (BoW) representation method is introduced which is different from the typical histogram-based approach. An image is converted to a document and then to a binary signature to have a control over retrieval speed by reducing the feature space. Furthermore converting an image into a document and to study how well it behaves with text processing techniques which simplify the CBIR has never been atempted. Therefore, this research forcuses on finding the best sub-image size which can be used in BoW. Sub-image is the main building block of this research and hence it is characterized by color, texture and shape features using BoW with the feature index and the index of the nearest cluster center. Then Random Indexing (RI) is introduced to CBIR by applying RI on the generated text file. The performance of the proposed approach is evaluated using three benchmark datasets for quality, speed and robustness which confirms that the proposed approach has a high potential to retrieve correct images which in turn can be extended for a large collection. System performance is compared with existing systems in the literature and the results prove that our approach has superior performance over the other systems.