Color template-based offline handwritten character extraction method

Abstract

The present invention provides a color template-based offline handwritten character extraction method. The method includes the following steps that: S11, a template image of a writing area is obtained from a preset template containing color squares and auxiliary lines; S12, black channels and color channels are extracted from the template image according to the characteristics of characters and borders, wherein the black channels are used for storing character handwriting features, and the color channels are used for storing the border characteristics of the preset frame; S13, binarization processing is performed on the black channels and the color channels, so that a binarized image is obtained, and stray points are removed from the binarized image, so that a standard image can be obtained; and S14, in the standard image, the image of the black channels is segmented according to the preset grids of the color channels, and black characteristic images can be obtained after segmentation, and the black characteristic images are adopted as extracted handwritten characters.

Description

technical field [0001] The invention belongs to the technical field of recognizing off-line handwritten characters by using computers, smart phones, tablet computers, cameras and other equipment in image recognition, and in particular relates to an off-line handwritten character extraction method based on a color template. Background technique [0002] In the prior art, an off-line handwriting recognition system generally includes three main steps: character segmentation, feature extraction, and character recognition, wherein character segmentation is a key part of handwriting recognition. To accurately segment the connected characters, the following methods have been proposed in recent decades of research: one is based on the interval and peak method, which uses the size of the vertical statistical value to segment characters; The second is a search-based segmentation method, which uses pre-trained categories to find matching characters and perform segmentation; the third m...

AI Technical Summary

Problems solved by technology

However, none of these methods can completely and accurately solve the segmentation problem of offline handwritten characters, let alone extract characters effectively and accurately.

Images