PROCESS FOR EXTRACTING INDIVIDUAL CHARACTERS FROM RASTER IMAGES OF A READ-IN HANDWRITTEN OR TYPED SERIES OF CHARACTERS IN FREE DISTRIBUTION

摘要

A separation classifier is used to calculate image-model-specific separation values for a separation image (PTR) of standard width (bstd) per column and determine the column with the maximum separation value as the right-hand separation column (tr). From the right-hand separation column (tr) predetermined by the separation classifier, a separation path is sought, starting with the upper image point on the separation gap which is formed within a separation range lying to each side of the separation gap if there is no white gap, partly by contour tracing and, if no white path can be found by contour tracing, partly by compulsory separation along the separation gap.