A Recognition Method for Skewed Four-State Bar Codes by Using Raster-Scanned Binary Images and Connected Components

A method for reading a four-state bar code, using raster-scanned images is described. Documents are pre- bar coded with one-dimensional four-state bar code that has the free location and no more than 5 degrees inclination. Using raster-scanned images captured by optical character recognition (OCR) equipment, the location and orientation of this bar code is detected and decoded to 20 data characters. The location detecting is done by labeling the binary image runs and merging the connected components of a code. The decoding is done by classifying the connected components based on their shape and location in the code line. Experiments demonsrrated that the rejection rate, error rate, and processing speed are suficient for practical application.