Accelerating a TV based JPEG decompression algorithm with Cuda
Institute of Mathematics and Scientific Computing, Karl-Franzens University of Graz
Karl-Franzens University of Graz, 2011
@article{holler2011accelerating,
title={Accelerating a TV based JPEG decompression algorithm with Cuda},
author={Holler, M.},
year={2011}
}
In previous works, we have have developed a mathematical model for artifact-free decompression of JPEG images. There, the problem of finding an artifact-free decompression for a given JPEG compressed image is related to a convex minimization problem. We use a primal-dual algorithm to solve this problem, for which we have developed a Matlab and C++ implementation. Since calculation time is crucial for the practical application of this model, we want to use the CUDA programming language in order to implement our code for graphic processor units (GPUs) for a further decrease of calculation time. On the following we will address this issue. We start by giving a short introduction to the JPEG compression standard [4], explaining the basic principles of our model and introducing a primal-dual algorithm used to solve the minimization problem [1]. For a more detailed mathematical treatment of the problem we refer to [2],[3]. In the second section, we describe the CUDA implementation of the algorithm. The last section compares calculation times for different implementations and images.
November 4, 2011 by hgpu
