PARALLEL SCHEME FOR MULTI-PASS QMAX AND MULTI-RATE QMIN ON GPU
3
Author(s):
SUDARSHAN. E, CH. SATYANARAYANA, C. SHOBA BINDU
Vol - 8, Issue- 9 ,
Page(s) : 42 - 55
(2017 )
DOI : https://doi.org/10.32804/IRJMST
Get Index Page
Abstract
Maximum Quantization Descent (MQD) uses a parallel multi-pass Qmax approach to search the matrix's most considerable wavelet coefficient value to encode independently along with the corresponding MQD node, as described in the first part of this article. In that connection, the generic serial comparison-based search algorithm has stopped processing acceleration because it could raise the complexity up to O(〖log〗_2 n)times, in that same work the parallel reduction-based search algorithm is could be marked at O(n^2 )times only. The second part of the article describes how to control a bit rate at a specified level using the parallel multi-level Qmin algorithm. This algorithm encompasses the encoding process 24% faster and completes the 27% faster-decoding process than BCWT without sacrificing the PSNR value.
|