Recognition of isolated handprinted characters

Handprinted characters are of unequal complexity and a common description of all alphabet symbols seems therefore unobtainable. However, letters which confuse human beings and man-made OCR systems usually have approximately the same appearance and may therefore be modeled jointly. We part the set of bitmaps into types, where each type has its unique feature space. The bitmaps belonging to some types are modeled independently from bitmaps belonging to other types. The feature vector of a bitmap initially constitutes a lossy representation of the contour(s) of the bitmap. The initial feature space is usually too large but can be reduced automatically by the use of a predictive code length or predictive error criterion