Please use this identifier to cite or link to this item:
|Title:||Implementação de um algoritmo de compressão de imagens baseado em recorrência de padrões multiescalas em CPU multinúcleo e GPU massivamente paralelas|
|Other Titles:||Implementation of an image compression algorithm based on multiscale recurrent patterns in multi-core CPU and GPU massively parallel|
|Keywords:||Compressão de imagem; recorrência de padrões multiescala; unidade de processamento gráfico; algoritmo; sistema de telecomunicação; Imaging compression; multiscale recurrent patterns; graphics processing unit; algorithm; telecomunication systems|
|Abstract:||The method of signal compression MMP (Multi-dimensional Multiscale Parser) is based on recurrency of multiscale patterns, where each block of the input signal is approximated by an adaptive dictionary, which is updated with expanded and compressed versions of concatenation of a previously coded block. This algorithm is structured by a binary segmentation tree. This class of algorithms has the characteristics of a high consumption of computational resources. With the advent of techniques and software libraries for the parallelization of applications on CPUs and GPUs, we identified the opportunity to investigate the possibility of acceleration of this class of algorithms.|
|Appears in Collections:||TEDE sem arquivo|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.