An Orthogonal Matching Pursuit Algorithm for Image Denoising on the Cell Broadband Engine

Patch-based approaches in imaging require heavy computations on many small sub-blocks of images but are easily parallelizable since usually different sub-blocks can be treated independently. In order to make these approaches useful in practical applications efficient algorithms and implementations are required. Newer architectures like the Cell Broadband Engine Architecture (CBEA) make it even possible to come close to real-time performance for moderate image sizes. In this article we present performance results for image denoising on the CBEA. The image denoising is done by finding sparse representations of signals from a given overcomplete dictionary and assuming that noise cannot be represented sparsely. We compare our results with a standard multicore implementation and show the gain of the CBEA.

[1]  Pierre Kornprobst,et al.  Mathematical problems in image processing - partial differential equations and the calculus of variations , 2010, Applied mathematical sciences.

[2]  Michael Elad,et al.  Submitted to Ieee Transactions on Image Processing Image Decomposition via the Combination of Sparse Representations and a Variational Approach , 2022 .

[3]  Harald Köstler,et al.  A parallel K-SVD implementation for CT image denoising , 2009 .

[4]  A. Bruckstein,et al.  On the uniqueness of overcomplete dictionaries, and a practical way to retrieve them , 2006 .

[5]  Joel A. Tropp,et al.  Topics in sparse approximation , 2004 .

[6]  Michael Elad,et al.  E-cient Implementation of the K-SVD Algorithm and the Batch-OMP Method , 2008 .

[7]  J. Hornegger,et al.  Separate CT-reconstruction for 3D wavelet based noise reduction using correlation analysis , 2007, 2007 IEEE Nuclear Science Symposium Conference Record.

[8]  Michael Gschwind The Cell Broadband Engine: Exploiting Multiple Levels of Parallelism in a Chip Multiprocessor , 2007, International Journal of Parallel Programming.

[9]  Rainer Raupach,et al.  Wavelet Based Noise Reduction by Identification of Correlations , 2006, DAGM-Symposium.

[10]  S. Mallat,et al.  Adaptive greedy approximations , 1997 .

[11]  Ulrich Rüde,et al.  Nonlinear Diffusion vs. Wavelet Based Noise Reduction in CT Using Correlation Analysis , 2007, VMV.

[12]  J. Craggs Applied Mathematical Sciences , 1973 .

[13]  Michael Elad,et al.  Image Denoising Via Sparse and Redundant Representations Over Learned Dictionaries , 2006, IEEE Transactions on Image Processing.

[14]  Michael Elad,et al.  Optimally sparse representation in general (nonorthogonal) dictionaries via ℓ1 minimization , 2003, Proceedings of the National Academy of Sciences of the United States of America.