An efficient approach towards building CBIR-based search engine for embedded computing board. (28th October 2022)

• Record Type: Journal Article
• Authors: Vasudevan, Shriram K.; Priyadarsini, P.L.K.; Sundaram, R.M.D.
• Is Part Of: International journal of advanced intelligence paradigms. Volume 23:Number 3/4(2022)
• Page Start: 318
• Page End: 331
• Publication Date: 2022-10-28
• Subjects: retrieval -- wavelet -- histogram -- texture -- OpenCV -- MATLAB -- region of interest; Artificial intelligence -- Periodicals; Machine theory -- Periodicals; Fuzzy logic -- Periodicals; 006.305
• Languages: English
• ISSNs: 1755-0386

Abstract

Investigating a picture gives us more information than how it is expressed through words. Image processing is such a field which is ever booming and handles ' n ' number of images. Thanks to technology we are able to store and retrieve such massive dataset of image-based data from anywhere. Search engines provide a way to link images and queries. They are searched using various factors like keyword, image dimension, texture etc. which is called as content-based image retrieval. In this search methodology, the input query image is analysed, and its properties or features are saved. Using the recorded features, other images are retrieved which match with the input image. But then again, searching by just name, colour or texture is not very efficient and so we have proposed a novel algorithm for the same. The proposed algorithm takes features like colour, texture, SURF, entropy, etc. and finds out how differently they work and what distinct results they produce when combined. Implementation of CBIR on Beagle board led us to some satisfactory results, which encouraged us to do further research.