The increasing demands of speed and performance in modern signal and image processing applications necessitate design and applications of massive parallel processors technology. Due to the availability of low-cost, high density, high speed Very Large Scale Integration (VLSI) devices and emerging computer aided design, there has been a dramatic worldwide growth in research and development efforts on mapping various signal and image processing applications onto such VLSI architectures. Modern image processing technology depends critically on the device and architectural innovations of the computing hardware. This paper attempts to survey some of the VLSI architectures that have already been reported for computation of Fast Fourier Transform (FFT) in signal and image processing applications. Primarily, pipeline-based, parallel architecture is discussed and parallel-pipeline FFT processor architectures are also touched upon. In the end, the paper also discusses the importance of using CORDIC algorithms for design and implementation of efficient VLSI architectures for the said computing method.

Signal and image processing encompasses a wide variety of mathematical and algorithmic techniques (Henry et al., 1974; Erling and Despain, 1984; and Ray, 1998). Most image processing algorithms are dominated by transform techniques, convolution/correlation filtering and some key linear algebraic methods. The dominating aspects in image processing requirements are essentially enormous throughput rates and huge amounts of data and memory. Fast Fourier Transform (FFT) is the most popular algorithm in digital signal processing. Pipelining, array processing and multiprocessing represent standard methods in computer organization which are commonly used for high-speed processing to reduce the inherent complexity in the design of large-scale multiprocessor arrays (Henry et al., 1974; and Ray, 1998).

The number of complex multiplication and addition operations required by the simple formsboth the Discrete Fourier Transform (DFT) and Inverse Discrete Fourier Transform (IDFT)are of order N², as there are N data points to calculate, each of which requires N complex arithmetic operations. FFT is the most popular algorithm in digital signal processing for the efficient and much faster computation of DFT. The DFT can be expressed as:

Telecommunications Journal, Very Large Scale Integration, VLSI, Fast Fourier Transform, FFT, Ddigital Signal Processing, Circuit Implementation, Image Processing, Pipeline Processors, Discrete Fourier Transform, DFT, Multimedia Technology, Signal Processing Perspectives.