Novel signal-processing techniques in barcode scanning

Abstract

Traditional barcode scanners use analog edge detectors to detect the boundaries of dark and light areas. As long as the edge strength exceeds a given threshold, the detected edge is considered to correspond to a bar/space boundary. Such a hard decision often leads to errors, especially in scanning barcodes that are poorly printed or scanned from a distance. An earlier study showed that much better results can be obtained by working directly with samples of the analog data. However, that method requires storing the entire sampled waveform of a scan line in memory, which is not cost effective. The method described in the article attempts to combine the performance advantages of processing the samples of the analog waveform with the low-cost advantage of the traditional methods. We still rely on edge-detection circuits, however, in addition to saving the edge time we also store information about the edge strength. In this way, we have been able to produce many of the benefits of the approach described in Joseph and Pavlidis (1993) with only a small increase in cost.