A distributed arithmetic based CORDIC algorithm and its use in the FPGA implementation of the 2-D IDCT
Yi Yang · Spectrum Research Repository (Concordia University) · 2002
The discrete cosine transform (DCT) based image compression techniques play an important role in today's digital applications. A video codec chip requires an integration of high-speed DCT and inverse DCT (IDCT) hardware units in a limited silicon space. This thesis presents a distributed arithmetic based CORDIC algorithm for the computation of the 1-D IDCT and an FPGA implementation of a cost-effective architecture for a 2-D IDCT processor using the proposed algorithm. The processor consisting of two 1-D IDCT cores, a transpose memory and a control logic block performs the 2-D IDCT computation by using the row-column decomposition approach. The basis of the proposed scheme is a combined use of the distributed arithmetic and the CORDIC algorithm in order to provide a small access time to the lookup tables and a reduced complexity for its architecture. In the proposed design, the deep pipeline structure of an existing CORDIC based architecture is replaced by much smaller DA-based ROM accumulators. In the proposed design, a bit-level digit-serial structure based on the redundant number system using an on-line algorithm is employed.