In designing this architecture, we include novel optimizations at the algorithmic, architecture, and circuit levels.Īn approach to image coding based on a fractal theory of iteratedĬontractive transformations defined piecewise is described. The design is also scalable so that larger images can be encoded faster by adding chips to the array. In this paper, we will present a parallel ASIC array architecture for use in fractal encoding that performs a full domain quad-tree search in near real-time for standard sized gray scale images. This has lead to a recent trend for designing parallel subsystems and including multimedia ASIC circuitry on general purpose CPUs. At the same time, VLSI area has become a much less important constraint in chip design due to better fabrication techniques and smaller micron technologies. The problem with fractal coding lies in the large amount of pixel block comparisons that are required, which makes fractal coding better suited toward parallel systems. Though fractal coding has been extensively optimized for speed, it is still not practical for real-time applications on most sequential machines. Fractal image coding is a compression technique with many promising features, but it has been primarily placed in the class of archival coding algorithms due to its computationally expensive encoding algorithm.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |