This paper derives new framework for content based image retrieval (CBIR) by deriving color features and texture features. This paper initially transforms the RGB color plane image into HSV color plane and derives the individual histograms on H, S and V planes. This paper derives rule based dynamic motif matrix (RDMM) on full texton index (FTi) index images of V-plane. The FTi overcomes the disadvantages and ambiguity issues in indexing textons of earlier methods. The RDMM overcomes the ambiguity issues arise especially when two or more pixels of the 2x2 grid exhibit the similar intensity levels and this reduces overall retrieval rate. This paper integrated the co-occurrence features derived on RDMM-FTi with histogram features of pure color model. The proposed framework is tested on popular color databases and also compared with state of art models of texton and motif based methods based on average precision rate (APR) and average recall rate(ARR). The experimental results indicate the efficacy of the proposed method over the existing methods.
This paper proposes an efficient method called tilted rectangle (TR) for detecting and correcting of slant angle of the manuscript Telugu words (MTW). Telugu language is one of India's common languages spoken by over 80 million individuals. The complex characters are attached with some extra marks known as “maatras” and “vatthus,” and it is challenging to detect slant angle. The proposed TR method initially performs preprocessing and identifies a connected component within the given Telugu manuscript word. Then, it estimates the slant angle of each connected component by deriving connected slant lines on the boundary of each connected component. After this process, the proposed TR method estimates the entire word's overall slant angle from the average of estimated slant angle and height of all connected components. The correction of the word's slant angle is done in the reverse direction by applying a simple shear transformation. With 1000 manuscript records of three different kinds, the algorithm is tested. Experimental findings indicate the efficacy of the approach proposed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.