Efficient Texture Image Retrieval Using Copulas in a Bayesian Framework

Abstract

Abstract-In this article, we investigate a novel joint statistical model for subband coefficient magnitudes of the Dual-Tree Complex Wavelet transform which is then coupled to a Bayesian framework for Content-Based Image Retrieval. The joint model allows to capture the association among transform coefficients of the same decomposition scale and different color channels. It further facilitates to incorporate recent research work on modeling marginal coefficient distributions. We demonstrate the applicability of the novel model in the context of color texture retrieval on four texture image databases and compare retrieval performance to a collection of state-of-the-art approaches in the field. Our experiments further include a thorough computational analysis of the main building blocks, runtime measurements and an analysis of storage requirements. Eventually, we identify a model configuration with low storage requirements, competitive retrieval accuracy and a runtime behavior which enables the deployment even on large image databases.

Gaussian Copula Multivariate Modeling for Texture Image Retrieval Using Wavelet Transforms

“…Even if probabilistic approaches can lead to a significant gain in retrieval accuracy, the drawback is its computationally expensive complexity. The optimum selection rule (2) could be high computationally expensive and a reduction of data size is required to decrease this complexity [26], [36]. However, this comes at the expense of the retrieval accuracy.…”

“…We consider a wavelet subband of a query image presented by observation We propose to analyze the computational complexity as a function of d , and e, so all involved numerical functions have complexity Ο 1 (they do not depend on d , and e). We tried to perform the same optimization described in [36] consisting of precomputing of matrix inversions or determinants when image database is indexing.…”

“…This MC procedure is computationally expensive and thus the KLD closed form, when it exists, is to be preferred [9]. Independently of our work, Kwitt et al recently took a similar approach where they introduced several copula-based models to characterize the association between subband coefficients from color channels at the same decomposition scale, in the context of color texture retrieval [36].…”

Texture Classification by Bit-plane Multifractal Spectrum and Bit-plane Barycentric Coordinates of Wavelet Coefficients Based on SVD

“…Despite lots of efforts in last decades, it still remains a challenging problem to model textures efficiently. Some available methods of texture classification and retrieval can be found in [1,2,3,4]. SVD is an important matrix theory and has been popularly employed in image processing, such as data compression [5], texture segmentation [6], and texture classification [7].…”

Contour Points based P2P Algorithm for Shape Matching and Image Retrieval

“…The color based approaches [4,5,6] utilize color features including color histogram, color set, color moment, color coherence vector, color correlogram, etc. The texture based methods [7,8,9] employ texture features including the gray level co-occurrence matrix, wavelet transform, Markov random field, local binary pattern, etc. The shape based techniques [10,11,12] adopt shape features including boundary chain code, Fourier descriptor, shape moments, etc.…”