VLSI Architecture Design of 9/7 Discrete Wavelet Transform for Image Processing

Abstract

In image processing, transform coding de-correlates images to pre-condition them for efficient compression. In this work, we propose VLSI architecture design of a hardware-efficient 9/7 Discrete Wavelet Transform (DWT). The architecture takes advantage of Canonical Signed Digit (CSD) and Distributed Arithmetic (DA) to represent and optimally distribute co-efficients to reduce the number of adder and shift registers. In addition, the co-efficient multiplication also exploits the horizontal and vertical redundancy in the architecture to reduce the hardware computational complexity. The result is a filter-based design, exploiting hardware path of architecture using CSD coefficients, which finds minimum realization. The proposed architecture is simulated using Verilog Hardware Description Language (HDL). A comparison with other architectures of 9/7 DWT shows a 18.75% reduction in hardware. The result is a hardware-efficient architecture, which provides a low-power solution for image and signal processing applications.