Matching pursuits video coding: dictionaries and fast implementation

Abstract

Abstract-Matching pursuits over a basis of separable Gabor functions has been demonstrated to outperform DCT methods for displaced frame difference coding for video compression. Unfortunately, apart from very low bit-rate applications, the algorithm involves an extremely high computational load. This paper contains original contribution to the issues of dictionary selection and fast implementation for matching pursuits video coding. First, it is shown that the PSNR performance of existing matching pursuits codecs can be improved and the implementation cost reduced by a better selection of dictionary functions. Secondly, dictionary factorization is put forward to further reduce implementation costs. A reduction of the computational load by a factor of 20 is achieved compared to implementations reported to date. For a majority of test conditions, this reduction is supplemented by an improvement in reconstruction quality. Finally, a pruned full-search algorithm is introduced, which offers significant quality gains compared to the better-known heuristic fast-search algorithm, while keeping the computational cost low.Index Terms-Displaced frame difference, low-complexity algorithm, matching pursuit, pruned full search, video coding.

A Geometrical Study of Matching Pursuit Parametrization

“…17 The interested reader will find a proof of this basic differential geometry result in the companion Technical Report [31].…”

“…An orthogonal 1-D or 2-D wavelet basis is also a trivial example of such a discretization even if in that case MP is not required to find signal coefficients; a simple wavelet decomposition is computationally more efficient. Works that do not directly rely on a prototype function either approximate such a parametrized dictionary based on computationally efficient cascades of filters [16][17][18], or attempt to adapt a set of parametrized dictionary elements to a set of training signal samples based on vector quantization techniques [19,20]. Thus, most earlier works define their dictionary by discretizing, directly or indirectly, the parameters of a prototype function.…”

Efficient Scalable Video Coding Based on Matching Pursuits

“…Many approaches have been proposed to simplify the complex encoding stage. One approach approximates the codewords of a dictionary with a linear combination of simpler codewords so that the computation is easier (Redmill et al, 1998;Czerepi´nski et al, 2000;Neff & Zakhor, 2002;Vleeschouwer and Macq, 1999;Jeon & Oh, 2003). This technique can be further developed by combining the inner product calculation and the atom finding components (Lin et al, 2007).…”

“…The traditional dictionary design for reducing complexity is the 2-D separable dictionary (Neff & Zakhor, 1997). To further reduce the complexity of this dictionary, another approach has been proposed to approximate it with a low cost factorized separable dictionary in which larger basis functions in the 2-D separable dictionary are represented as a successive convolution of short basis functions (Vleeschouwer & Macq, 1999;Czerepi´nski et al, 2000). Although the complexity is drastically reduced, the disadvantage of this approach is that the dictionary is restricted within a 2-D separable set; thus, the design of the dictionary may not be optimal in terms of coding performance.…”

SNR scalability based on bitplane coding of matching pursuit atoms at low bit rates: fine-grained and two-layer

“…The encoder of an MP-based video codec is extremely intensive in computation. Several methods have been proposed to reduce the computational load by representing the bases in a dictionary as a combination of simpler bases whose inner products can be obtained with less computational complexity [12], [22], [3]. It was shown in [15], [23] that the performance of a nonscalable MP codec can be improved by in-the-loop quantization where quantization noise is included in the MP algorithm.…”