Chinese character font automatic generation method based on skeleton guide transmission network

Abstract

The invention discloses a Chinese character font automatic generation method based on a skeleton guide transmission network. The method comprises the steps of data preprocessing, font skeleton synthesis network construction, font style correction network construction, discriminator loss function construction, and Chinese character font automatic generation. Training is carried out by adopting a small number of stroke samples, fonts are decomposed into a font skeleton synthesis network and a font style correction network, a good stroke structure is achieved, and realistic fonts are synthesized; an expansion convolution DC is introduced into an encoder, an attention module ECA is introduced into a decoder, a stroke space structure is reserved, a vivid synthetic font skeleton is obtained, stroke detail information loss is reduced, and stroke feature representation capability is improved; meanwhile, a style correction network based on a multi-scale feature fusion module F is adopted, strokes or fonts of predicted fonts are refined, and important features are selected from different resolutions. In the Chinese character generation process, any handwritten font can be selected, the requirement limitation on input is reduced, and convenience is provided for Chinese character font generation.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a method for automatically generating Chinese character fonts. [0002] technical background [0003] Calligraphy is an art carried on Chinese characters and expressed in strokes, gestures and composition. The art of Chinese calligraphy is an art with a history of more than 3,000 years. It has an important position in the history of world art and embodies the Chinese way of thinking and aesthetic taste [1]. Calligraphy works of past dynasties are one of the most representative symbols of Chinese national culture. They not only record historical culture, historical events, and historical environment in content, but also have distinctive historical and personal characteristics in writing form. For example, Su Shi’s "Red Cliff Fu", Ouyang Xun's "Zhong Ni Meng Lian Tie", Wang Xizhi's "Lanting Collection Preface". Today, the purpose of learning traditional calligr...

AI Technical Summary

Problems solved by technology

However, since Chinese font generation is not a fault-tolerant task, it is an uncontrolled and unpredictable process, any blurring or ghosting, artifacts may lead to failure of generated fonts and result in poor quality of synthesized font results

[0005] Currently, most commercial typeface generation is focused on professional typeface design, which is time-consuming and labor-intensive work

Due to the following three reasons, it is still a challenging task for ordinary people to generate fonts with calligrapher handwriting: 1) Chinese characters are complex in structure and font style, and different people have different handwriting styles and strokes shape

2) The vocabulary of Chinese characters is very large, and it is difficult for people to correctly write Chinese characters with a consistent style with such a huge vocabulary

Its performance largely depends on the fine adjustment of each glyph. The complex and flexible Chinese character structure is impossible for ordinary customers to perfectly extract strokes or glyphs. Therefore, it is necessary to quickly build personalized Chinese handwritten fonts suitable for ordinary people. not possible

Images