Apparatus for performing character segmentation using slant histograms

Abstract

An apparatus for performing character segmentation of digitized handwritten character data utilizes a vertical histogram processing unit to identify primary cuts to be made in the character data based on a vertical histogram. Character blocks generated after the primary cuts are analyzed to determine if the character blocks contain multiple characters. A slant histogram processing unit then generates a set of slant histograms for each of the character blocks to identify minima indicative of segmentation points. The slant histograms are evaluated by an evaluation processing unit to determine segmentation points for kerned characters based on zero-value minima. After segmenting the kerned characters to generated new character blocks, the evaluation processing unit evaluates the minima in accordance with a set of predetermined rules to identify further segmentation points for touching characters.

Dependencies are inferred from the claim text.