Multi-style dynamic word grouping method based on generative adversarial network

Abstract

The invention discloses a multi-style dynamic character grouping method based on a generative adversarial network, which comprises the following steps that: Chinese characters are expressed as an IDS sequence, and the IDS sequence comprises a plurality of characters forming substructures of the Chinese characters and IDC characters representing layout information of the substructures; constructing a font vector model of each substructure, and performing vector combination on the font vector models of the substructures according to layout information of the substructures presented by the IDC characters to obtain font vector models of Chinese characters; and performing style optimization processing on the font image corresponding to the font vector model of the Chinese character by using a style optimization model constructed based on the generative adversarial network. According to the method, preliminary dynamic character combination is carried out through a graphic method, and dynamic character combination results of various styles of fonts are optimized based on the adversarial network, so that a better font combination effect is realized.

Description

technical field [0001] The invention belongs to the field of computer-aided design, and in particular relates to a multi-style dynamic word grouping method based on a generative confrontation network. Background technique [0002] It has always been an important project in the field of Chinese character design to realize the expansion of the character set for Chinese characters of a specific style. In recent years, deep learning techniques, especially style transfer methods based on variational autoencoder VAE and generative adversarial network GAN, have made some progress in the generation of Chinese character glyphs. Some of the representative works are: Zhang Y, Zhang Y, Cai W.Separating style and content for generalized style transfer[C] / / Proceedings of the IEEE conference on computer vision and pattern recognition.2018:8447-8455.Xie Y , Chen X, Sun L, et al. DG-Font: Deformable Generative Networks for Unsupervised Font Generation[C] / / Proceedings of theIEEE / CVF Conferen...

AI Technical Summary

Problems solved by technology

[0007] All in all, the current method that only uses deep learning is mainly for modern Chinese characters in the conventional character set, and cannot generate special Chinese characters outside the character set very well, and there are certain limitations in the results of dynamic group character generation using only graphics methods

There is still a lack of research on combining the two for multi-style dynamic optimization.

Images