Serial Number Identification with Machine Vision
Danny Persaud, Adam Patt, Calin Miculescu · The Atrium (University of Guelph) · 2008
In this paper, DAC consulting presents a new method of acquiring accurate optical character recognition (OCR) results through the use of machine vision. The design consists of implementing a new software called Intellect 1.5, which allows us to preprocess the image by applying a layered aggregate of filters. We then apply an edge detection tool called a fiducial and finally an OCR soft-sensor. Once these layers are applied in order, the software is taught the series of characters to which it must recognize.