Improving the performance of 2D Discrete Wavelet Transform using data-level parallelism

The JPEG2000 standard uses the 2D Discrete Wavelet Transform (2D DWT), while the JPEG standard uses the 2D Discrete Cosine Transform (DCT). However, the 2D DWT has higher computational requirements than the 2D DCT and consumes a significant part of the total JPEG2000 encoding time. One way to improve the performance of the 2D DWT is using parallel techniques on an SIMD-enhanced architecture. In this paper, we apply data-level parallelism technique to exploit available parallelism of the 2D DWT. We focus on the two algorithms to traverse an image to implement the 2D Discrete Wavelet Transform (DWT), namely Row-Column Wavelet Transform (RCWT) and line-Based Wavelet Transform (IBWT). Our experimental results show that the SIMD implementation of the IBWT algorithm is more complicated than the SIMD implementation of the RCWT algorithm, while the former algorithm is 1.60 times faster than the latter algorithm for an image of size 4096 × 4096.