Font style migration method and system based on multi-task generative adversarial network

Abstract

The invention provides a font style migration method and system based on a multi-task generative adversarial network, and the principle is that a cross-domain migration condition is added as the input of a generator, a plurality of font samples are input at the same time, the conversion from different font styles is supported, the problem of one-to-one font style migration is solved, and therefore, migration among a plurality of font styles is realized; according to the invention, when the generative adversarial network model is constructed, a feature attention module is added in a generator of the generative adversarial network, the feature learning ability of the generator in the generative adversarial network is improved, and the image generation quality is improved by improving the capture ability of local detail features of the font pictures.

Description

technical field [0001] The invention belongs to the field of computer vision and image processing, and in particular relates to a font style migration method and system based on a multi-task and feature attention mechanism generating confrontation network. Background technique [0002] The design of font style is an extremely difficult job, and it is still considered the prerogative of professional designers, and the existing software systems do not have the creativity of professional designers. However, most commercial font products are actually hand-designed with specific requirements for certain glyph attributes, such as italic, serif, cursive, width, angle, etc. [0003] The Chinese font style design is particularly difficult. According to statistics, in Chinese characters, 1000 commonly used Chinese characters can cover about 92% of written materials, 2000 Chinese characters can cover more than 98%, and 3000 characters can cover 99%. Therefore, at least about 2800 sing...

AI Technical Summary

Problems solved by technology

[0008] In view of the above problems, the present invention provides a font style migration method and system based on a multi-task and feature attention mechanism, which can overcome the problem that the font migration model in the prior art only supports single font conversion, and improve the font style. Migration effect, realizing font style migration with a small number of target fonts

Images