Computational challenges in frames-based signal and image processing

Darian Onchis

26 March 2014

Nowadays, we are experiencing an impressive level of sophistication in the development of algorithms for signal and image processing in the same time with an exponential increase of computing power. In order to cope with the demands of high-tech applications, besides the use of more powerful computers it is essential to design problem adapted algorithms for analyzing and synthesizing signals and images. The frame-based approach is using a superpositions of a non-orthogonal collections of building blocks to realize constructively the adaptive linear representation of nD objects under constraints such as numerical stability or the efficient use of memory. In this talk, we will overview recent advances in the frames-based signal and image processing and we will answer questions concerning precision, condition numbers, computational costs or parallelization requirements. Also, we will outline future directions of study and open problems.