摘要：In order to accelerate the Discrete Wavelet Transform DWT, we have implemented and
compared the lifting "Le Gall5/3" and "Cohen-Daubechies-Feauveau9/7" (CDF9/7) algorithms
on a low cost NVIDIA’s GPU. The suggested implementation is realized in MatLab using the
in-house parallel computation toolbox (PCT). Our experimental results indicate, that the
speedup is proportional to the image size until it attains a maximum at 20482
pixels, beyond
these values the curve decreases. The performance with GPU enhances above a factor of 2~3
compared with CPU.
关键词：Discrete Wavelet Transform (DWT); Le Gall5/3; Cohen-Daubechies-Feauveau9/7; GPU;;
CPU; MatLab.
