Image processing using a universal nonlinear cell

Abstract

Silicon implementation of many image processing algorithms has been hindered in the past due to their complexity and computational volume. This paper discusses two such algorithms, namely the 'Detection of line segments via Hough transform' and 'Backprojection in CT image reconstruction'. To gain a significant speed advantage, we present a multi-function cell for performing one of four nonlinear operations: (1) square-root, (2) reciprocal, (3) sine/cosine, and (4) arctangent/spl minus/all realized in a single chip, available on a selectable basis, and outputting a new result every two clock cycles. The design and test results of a 24 bit and a 16 bit four-function "two cycle" VLSI chip, both fabricated in 2.0 micron CMOS technology, are described. Also discussed is a new "one-cycle" architecture, which can potentially double the throughput of the chip. Using this nonlinear cell an application level Hough transform module is developed. The Hough module can deliver parameters of the line in the image plane every clock cycle.<>