ONE-DIMENSIONAL DIGITAL PROCESSING OF IMAGES FOR STRAIGHT-LINE DETECTION

摘要

The detection of straight lines in images is a common requirement in the recognition of some patterns. Several approaches have been in the use for many years but are computationally intensive. This paper presents a simple algorithm which can make use of the fast fourier transformer (FFT) for rapid computation and has the added feature of being able to detect straight lines of a specified length. The method herein suggests that simply unlacing the image raster before matched filtering can reduce the search range because of the restricted periodicities in the one-dimensional unlaced signal. These matched filters can be applied in the frequency domain through the use of the FFT. This note discusses the algorithm and presents example results.