Method and device for touching character segmentation in character recognition

Abstract

The present invention discloses a method and a device for touching character segmentation in character recognition, and belongs to the character recognition field. The method comprises the steps of carrying out preprocessing to obtain connected domains, an average character width and an average character height of a row image; carrying out connected domain analysis, marking touching connected domains, carrying out stroke extraction for selected touching connected domains, carrying out segmentation point detection of the touching connected domains to obtain pre-segmentation points, and saving character blocks for non-touching connected domains; merging extracted strokes according to the pre-segmentation points to obtain the character blocks; saving the character blocks and turning to a next connected domain for carrying out touching determination, and outputting a character block sequence after all connected domains having been traversed; and merging the character blocks according to reference information and outputting recognition result. The method and the device provided in the invention merge the strokes according to the pre-segmentation points to obtain the character blocks, guaranteeing that segmentation points in a larger scope can be detected, and take the mode that contour information is used to predetect the segmentation points as a parameter in merging, avoiding merging mistakes caused by merging correct segmentation points.

Description

technical field [0001] The invention belongs to the field of character recognition, and relates to a character segmentation method and device, in particular to a character segmentation method and device for character recognition. Background technique [0002] In the text recognition process, character segmentation is a very important part. Text recognition, especially offline handwritten text recognition, generally adopts an over-segmentation method, that is, first detects multiple possible segmentation points, and then merges the segmentation points through geometric, recognition or semantic information to search for the optimal Split paths. [0003] There are many ways to detect segmentation points. The detection of segmentation points in printed text recognition is relatively simple, but in handwritten text recognition, the situation of handwriting adhesion is more complicated. There are many false peaks and troughs in the outline, and the real ones There may be adhesio...

AI Technical Summary

Problems solved by technology

[0003] There are many ways to detect segmentation points. The detection of segmentation points in printed text recognition is relatively simple, but in handwritten text recognition, the situation of handwriting adhesion is more complicated. There are many false peaks and troughs in the outline, and the real ones There may be adhesion between the upper contour and the lower contour on the segmentation point, which makes the contour change relatively smooth, so it is not reliable to determine the segmentation point only by contour information

[0004] At present, there are mainly two methods to find the segmentation points of offline handwritten characters. The first method is based on the connected domain, and judges the connected domain that may contain glued characters through the recognition effect or geometric information, and then finds out in the suspicious connected domain. Segmentation using information such as contours. The accuracy of this method depends on the recognition effect and geometric information. It is not reliable for handwritten character recognition, and segmentation based on contour information cannot determine the segmentation point.

Another method is based on stroke segmentation. The strokes in the segmented offline handwritten characters are merged into stroke blocks, but the overlapping strokes may overlap when merging, so that the strokes in the two connected domains overlap due to their positions. was merged, resulting in inaccurate mergers

Images