Implementation of Multiplication-Free Linear Transforms on a Programmable Processor

Abstract

Many signal processing applications such as image transforms [27], error correction/detection involve matrix multiplication of the form Y = A * X, where X and Y are the input and the output vectors and A is the transformation matrix whose elements are 1,-1 and 0. This chapter presents optimized code generation of these transforms targeted to both register rich RISC architectures such as ARM7TDMI [6] and single register, accumulator based DSP architectures such as TMS320C2x [95] and TMS320C5x [96].