Sample-based texture synthesis method

Abstract

The invention relates to a sample-based texture synthesis method, which comprises the following steps: 1) randomly selecting size-specific sample blocks from input texture to form a sample block sequence; 2) randomly selecting a block B1 from the sample block sequence and placing the block to an output image; 3) at the place of each block to be synthesized, searching a texture block B2 meeting matching conditions from the input sample block sequence and placing the texture block B2 to a synthesis image to enable the B2 to have a certain overlap region with the B1; 4) calculating multi-order error surface of the overlap region of the B1 and B2; 5) finding an optimum segmentation path from the error surface calculated in the step 4) as the boundary of the new texture block B2, and splicing the B2 to the synthesis image; 6) after splicing, introducing Poisson fusion to optimize boundary un-matched regions around the segmentation path; and 7) carrying out the steps 1)-6) until finishing texture synthesis.

Description

technical field [0001] The invention relates to a texture synthesis method, in particular to a texture synthesis method suitable for synthesizing structural textures and natural random textures. Background technique [0002] With the continuous development and progress of science and technology, the popularization of computer networks and the increase of people's demand for digital image information, texture synthesis technology has attracted more and more attention. The generation of scenes and the rendering of realistic images have shown broad application prospects, and it is one of the research hotspots in the fields of computer graphics, computer vision and image processing. Texture synthesis is based on given small area texture samples, according to the geometry of the surface, flatten to generate texture images of any size, which are visually similar and continuous. Texture synthesis methods mainly include two categories: one is the feature matching method, which rega...

AI Technical Summary

Problems solved by technology

In the prior art, the block-based texture synthesis method is faster than the point-based texture synthesis method, and can maintain the consistency of the texture structure to a certain extent, through the block-by-block synthesis as follows instead of point-by-point, but still There was an issue where texture compositing images sometimes had too much repetition, resulting in boundary mismatches for many textures with strong structures and natural random textures

Images