Hidden Markov model for bar code denoising
Fadil Santosa, Dariusz Madej, Saša Krešić–Jurić · 2002
A bar code signal is decoded by identifying spikes in the signal which correspond to edges between bars and spaces in a bar code pattern. We show that such a signal can be modelled as a hidden Markov chain with eight states. The model is designed to label the spikes which correspond to bar code edges and those which correspond to noise, thus effectively denoising a bar code signal. The performance of the algorithm is compared to other existing decoders.