Method for automatically generating ink painting based on generative adversarial network GAN

Abstract

The invention discloses a method for automatically generating an ink painting based on a GAN (Generative Adversarial Network). The method comprises the following steps of 1), obtaining an ink paintingpicture data set, and carrying out the preprocessing of an ink painting picture; 2) decomposing the preprocessed ink and wash painting data set into different types of data sets; 3) denoising the data set by using a non-local mean denoising algorithm to obtain a feature data set; 4) establishing a generative adversarial network GAN by using the training data set, and determining the size of an input picture; 5) inputting the feature data set into a GAN (Generative Adversarial Network) for training to obtain a trained GAN neural network model; and 6) inputting the category label data into thetrained GAN neural network model, and automatically generating the ink and wash painting corresponding to the label. The method is advantaged in that problems of tedious manual operation and low creation efficiency in the traditional design method are solved.

Description

technical field [0001] The invention relates to the fields of deep learning and computer vision, in particular to a method for automatically generating ink paintings based on a generative confrontation network (GAN). Background technique [0002] In recent years, image research using deep neural networks has become a hotspot in machine learning and computer vision research. Generative adversarial networks, originally proposed by Ian Goodfellow, effectively encourage the output of the generator to be similar to the original data distribution by applying an adversarial loss to both the generator and the discriminator during the actual training process. GAN has achieved impressive results in image generation, image transfer, super-resolution and other generative tasks, and is widely used in various application scenarios. Using computers to generate images of various artistic styles is a hot issue in the field of digital images. The automatic generation technology of artistic s...

AI Technical Summary

Problems solved by technology

Ink painting has a special texture and aesthetic feeling. It pays attention to the combination of brushwork and ink, and requires special painting materials. Because of this, the traditional hand-painted ink animation is extremely complicated, time-consuming, and cannot be modified.

The use of existing production software to create ink-style paintings is also limited by the tools. The complexity is no less than that of manual work, and the effect is not ideal. There are certain limitations.

Images