Image generation system and method for generating image

Abstract

The invention discloses an image generation system and a method and includes an image database and a generation sub-system. The image database is used for storing local image sample information in classes and the correspondent local image units. The generation sub-system extracts geometric features from an input image, extracts each local image unit based on the geometric features, searches the local image sample the most similar to each local image unit from a training sub-system, and synthesizes local image units corresponding to each local image sample to an integral image. With the invention, an image with individuated features can be automatically generated based on a real image provided by a user with reduced calculation and improved image generation efficiency.

Description

technical field [0001] The invention relates to graphics processing technology, in particular to a portrait generating system and a method for generating portraits according to images. Background technique [0002] At present, with the rapid development of the Internet, the importance of online virtual life has become increasingly prominent. Compared with the characters in real life, online images have become an important form of personal presentation in cyberspace. However, very few people will use personal real photos as the image endorsement of the network environment, and often use cartoons to express virtual image positioning. At present, cartoon images are mostly developed by network content service providers, and users can choose from them. The advantage of this method is that the cost is low, and as a value-added part of the network service, it can be provided to users for free. The problem is that with the huge number of users, the cartoon images appear to be the ...

AI Technical Summary

Problems solved by technology

[0005] However, the main defect of the above method is that it uses short lines as the basic geometric elements to form a cartoon portrait. Often a portrait is composed of many short lines, and each short line needs to be extracted, analyzed and calculated, so the amount of calculation is large and the processing efficiency is low. Low

In addition, this method also requires a strict correspondence between the original photo and the portrait, so that only a limited cartoon style can be generated. Although it strives to be faithful to the photo, it is too rigid and difficult to expand the style, such as supporting exaggerated styles, etc.

Images