Preprocessing of Signals from Bar-Coded Silicon Using a Microprocessor

Bar code is a useful method of data entry in an automated environment. The method is very popular in normal label reading (paper label with black bars on white background) because of it simplicity. However, for materials where the contrast is not good, such as silicon, the signal processing becomes more complex. This paper describes a signal-processing technique developed to measure the widths of the bars and spaces and classify them according to width sizes based on a Nearest Neighbor (NN) approach. The classified data can then be decoded to give the equivalent alphanumeric representation of the code.