Personalized handwriting migration method and system based on collaborative stroke optimization

Abstract

The present invention provides a personalized handwriting migration method and system based on collaborative stroke optimization. The method includes: according to the average area ratio and average aspect ratio of the target Chinese character, deforming the size and proportion of the Chinese character to be input, so that the Chinese character to be input is aligned with the skeleton of the target Chinese character; the Chinese character after deformation processing is input into the target neural network, and the corresponding target Chinese character is output through the target neural network; wherein, the target neural network refers to A trained adversarial generative network for converting input Chinese characters into target font Chinese characters. In this way, it is possible to migrate any printed Chinese character to another printed or handwritten Chinese character by relying on a small amount of data sets, and even to achieve personalized handwritten font customization, and to quickly, accurately and realistically generate target fonts.

Description

technical field [0001] The invention relates to the technical fields of computer vision and image processing, in particular to a personalized handwriting migration method and system based on collaborative stroke optimization. Background technique [0002] Fonts are an important part of media content creativity, and are widely used in various visual communication designs in contemporary society, meeting the multi-directional and multi-level needs of mass media. The development of a set of Chinese fonts requires a lot of manpower and time. Different from the English alphabet (including uppercase and lowercase) that only contains 52 characters, at present, the lowest standard Chinese character encoding character set - GB2312-80 national standard code has selected a total of 6763 Chinese characters (including 3755 commonly used Chinese characters at the first level, and 3755 common Chinese characters at the second level. Second commonly used Chinese characters 3008). Therefore...

AI Technical Summary

Problems solved by technology

[0004] However, the existing methods mainly have the following two limitations: none of the existing font generation models based on image conversion considers the introduction of a process or module for optimizing Chinese character strokes in the model; most existing font generation methods rely on Learning from a large number of paired training samples (3000 pairs) does not greatly reduce time and labor costs, and is not very practical

Images