Intelligent image chart synthesis method based on graph convolutional network

Abstract

The invention discloses an intelligent image chart synthesis method based on a graph convolutional network. The method is characterized in that a deep neural network is adopted to generate a saliency graph of an input image, blue noise sampling is carried out based on the saliency graph, a triangulation grid is constructed based on sampling points, and a graph is constructed based on the triangulation grid. A reasonable chart position and size are predicted by using a graph convolutional network, a significant color of the target position is extracted, a color scheme with a large contrast with a background color is selected from a self-defined chart color library as a palette, and finally a visual chart is drawn according to data input by a user. Compared with the prior art, the method has the advantages that the harmonious and attractive visual chart can be automatically synthesized on the image, so that the method can be widely applied to scenes such as scene data visualization, augmented reality visualization and geographic data visualization, and, according to the image, the visual chart and corresponding data specified by a user, a visual chart which is reasonable in position and harmonious in color is rendered on the image.

Description

technical field [0001] The invention relates to the technical field of automatic graphic design, in particular to an image graph intelligent synthesis method based on a graph convolution network. Background technique [0002] In the field of graphic design, there are usually some design tasks with repetitive requirements, and automated graphic design can effectively help professional designers and ordinary users get rid of these repetitive tasks and focus on more creative designs. At present, there have been related works focusing on the automatic typesetting design of different elements, including poster design in which text is embedded in pictures, magazine content layout in which text paragraphs and pictures are independent of each other, floor plan design, etc., but no scholars have been found. The problem of image graph intelligent synthesis and its solution are proposed. The previous graphic design technical solutions mainly adopted a rule-driven approach, including p...

AI Technical Summary

Problems solved by technology

However, these methods still require manual participation. For example, layout templates and aesthetic rules rely on the design experience of domain experts and need to be provided by domain experts.

[0003] The graphic design method in the prior art is mainly based on the data-driven approach, that is, to construct a large number of data sets, and use the deep learning model for feature learning to achieve the effect of intelligent synthesis, but the collection and labeling of a large number of high-quality data sets is time-consuming. labor-intensive, inefficient

Images