Colour image retrieval using pattern co-occurrence matrices based on BTC and VQ

Abstract

Proposed is an effective feature for colour image retrieval based on block truncation coding (BTC) and vector quantisation (VQ). Each input colour image is decomposed into Y, Cb and Cr components. BTC is performed on the 4 × 4 Y blocks, obtaining a mean pair sequence and a bitplane sequence, and then they are quantised with the contrast pattern codebook and visual pattern codebook to obtain the contrast and visual pattern co-occurrence matrix. VQ is performed on the 4 × 4 Cb blocks and Cr blocks with the Cb codebook and Cr codebook, respectively, to obtain the colour pattern co-occurrence matrix. Retrieval simulation results show that, compared with two existing BTC-based features, the proposed feature can greatly improve retrieval performance.Introduction: Colour image retrieval has been an attractive research area for several decades. Early text-based image retrieval schemes utilised keywords for similar image retrieval, however they are impractical for modern image databases since the size of the databases is huge and different people may annotate the same image with different keywords. Unlike the traditional text-based image retrieval methods, content-based image retrieval (CBIR) methods index images based on their visual contents described by automatically extracted features, such as colour [1], texture [2], shape [3], and so on. Vector quantisation (VQ) and block truncation coding (BTC) are two typical block-based image coding schemes that can be used not only to efficiently compress images but also to extract compressed-domain features, and thus several VQbased [4-6] and BTC-based [7, 8] image retrieval methods have been proposed recently. Reference [4] extracted features directly from the codeword indices of the spatial-domain VQ compressed image. Reference [5] extracted features from the individual codebook generated from the image. Reference [6] extracted 12 DCT-domain vector quantisation index histograms based on the YCbCr colour space. Reference [7] extracted a block colour co-occurrence matrix (BCCM) and a block pattern histogram (BPH) from single bitplane BTC compressed data based on the RGB colour space. Reference [8] extracted three block pattern histograms (BPH) from three-bitplane BTC compressed data and three colour histograms (CH) based on the RGB colour space.In this Letter, we propose a new effective feature for colour image retrieval based on two pattern co-occurrence matrices generated from a BTC compressed Y image and VQ compressed Cb and Cr images, respectively.

Comparative analysis of wavelet transform algorithms for image compression

“…The continuous wavelet transform was established as another to short time Fourier transforms, to overcome the resolution problem [8]. The continuous wavelet transform is well-defined according [5].…”

“…[8]. The relative discrete subclass of the half plane consists of all the points (am, nam b) with (m, n) in 'Z'.…”

Query by Colour : Investigating the Efficacy of Query Paradigms for Visual Information Retrieval

“…Colour retrieval has been a major focus of content-based image retrieval since its inception (Kato 1991) and research into the development of effective retrieval approaches continues unabated despite a lack of empirical evidence regarding its efficacy as a retrieval mechanism (Hua 2010;Yu, Luo, and Lu 2011).…”

Color Texture Image Retrieval Based on Local Extrema Features and Riemannian Distance

“…The first BTC-based retrieval scheme for color images was proposed in [22] followed by some improvements a few years later [23,24]. Until very recently, one has been witnessing the evolution of BTC-based retrieval frameworks such as the ordered-dither BTC (ODBTC) [25,26], the error diffusion BTC (EDBTC) [27] and the dot-diffused BTC (DDBTC) [28].…”