Analysis and implementation of Discrete Wavelet Transform for compressing four-dimensional light field data

In this paper, we compare two different light field data compression schemes by applying a conventional 2-D Discrete Wavelet Transform (DWT) and the proposed 4-D DWT to the compression flow. The results suggest that 4-D DWT is better in terms of the compression rate and PSNR. The proposed flow is simple, efficient, and suitable for hardware implementation. We also analyze the memory bandwidth, internal buffer size, and latency of three different 4-D DWT hardware architectures. A 4-D DWT processor using the (2-D, 2-D) architecture is implemented, where the processor speed is about 6 times faster than its software counterpart.