A Contextual Postprocessing System for Error Detection and Correction in Character Recognition.

The paper is an examination of the effectiveness of various forms of contextual information in a postprocessing system for detection and correction of errors in words. Various algorithms using context are considered, from a dictionary algorithm which has available the maximum amount of information, to a set of contextual algorithms using binary n-gram statistics. The latter information differs from the usual n-gram letter statistics in that the probabilities are position-dependent and each is quantized to 1 or 0 depending upon whether or not it is nonzero. This type of information is extremely compact and the computation for error correction is orders of magnitude less than that required by the dictionary algorithm. The techniques described in the paper can allow relatively poor classifiers to become reliable systems by drastically cutting error rates with only modest reject rates. Experimental results are presented on the error, correction, and reject rates that are achievable as a function of the type of contextual information employed, and the size of the data base from which this information is obtained. Author