Thepade's Ternary Block Truncation Coding with various degrees and Color Spaces for Content Based Image Retrieval
Sudeep D. Thepade, Hemant B. Kekre, Anil Tukaram Lohar, Nmims Mumbai · 2013
Proficient Content Based Image Retrieval system is a much needed system for plenty of image retrieval supporting applications, which retrieve images more rapidly from giant image databases. Block Truncation Coding (BTC) is one of the prominent and simple techniques used most of the time. This method is built on color properties of an image for image retrieval. In this paper a novel Static Thepade's Ternary Block Truncation Coding with degree values (varied as n=1 to 5) and Dynamic Thepade's Ternary Block Truncation Coding are proposed. Total 36 variations of proposed image retrieval methods, using five degree values for Static Thepade's Block Truncation Coding and Dynamic Thepade's Block Truncation Coding with six color spaces (RGB, LUV, YCgCb, YCbCr, YUV and YIQ) are experimented on image database having 1000 images in 10 different categories. Experimental results have shown better performance in YCbCr color space followed by YUV. The best image retrieval is by Dynamic Thepade's Ternary Block Truncation Coding in YCbCr color space.