Algoritmo de casamento de padrões aplicado na estimação de movimento em compressão de video

AUTOR(ES)
DATA DE PUBLICAÇÃO

2003

RESUMO

With the application of motion estimation and compensation techniques in video sequences, we can reduce the number of bits needed to encode the video sequence and to increase the quality of the reconstructed video signal. On the other hand, the high computational cost of the process demands high performance of the encoder. This thesis proposes a new motion estimation algorithm that reconciles processing quality and computational speed. Coding problems and severa1estimation algorithms are analyzed with emphasis on the quality and processing speed issues. A new motion estimation algorithm, ca11edMSEA (a modification of the SEA algorithm) is presented. The algorithm k:nown as SEA (Successive Elimination Algorithm) obtains the optimum motion vector by perfonning a pre-selection of macroblocks in the search area. The SEA algorithm is modified by introducing new discard criteria applied in increasing order of complexity and by reordering the reading of macroblocks inside of the search area. An expressive improvement in processing speed is obtained and the optimum solution of the motion vector estimation, also achieved by exhaustive search and SEA, is preserved. The algorithm is implemented as part of an MPEG-2 encoder and it is shown that MSEA achieves adequate rate x distortion curves at a smaller computationalcost than its competitors. Also, the possibility is shown of mixing MSEA with other estimation algorithms, improving the rate x distortion x processing time curves with respect to other estimation methods used in practice (two dimensionallogarithm search and hierarchical search)

ASSUNTO(S)

compressão de imagens videodigital transmissão de imagem

ACESSO AO ARTIGO

Documentos Relacionados