Seamless image cloning method based on matrix decomposition

Abstract

The invention discloses a seamless image cloning method based on matrix decomposition. According to the method, an original linear system equation is built in the gradient domain method; matrix decomposition is carried out on a coefficient matrix of an original linear system to form a banded diagonal matrix in a decomposed mode, and the banded diagonal matrix is decomposed into a plurality of small banded diagonal matrixes; according to each small banded diagonal matrix, a small system equation is built, and the small system equation is solved through different threads of a GPU so as to obtain an original pixel compensation value of each pixel; the original pixel compensation values of all the pixels are substituted into the original linear system equation by being used as initial values of an original system so as to directly obtain a final pixel compensation value; furthermore, pixel values of all the pixels in a fusion area are obtained, and seamless cloning is completed. In the seamless image cloning method based on matrix decomposition, the data parallel processing capability of the GPU is fully used, operational speed is greatly increased, real-time image processing becomes possible, memory consumption is low, and the seamless image cloning method can be widely expanded to application programs.

Description

technical field [0001] The invention relates to the field of computer images, in particular to an image seamless cloning method based on matrix decomposition. Background technique [0002] In recent years, gradient domain methods have been widely used in the field of image processing, including intrinsic image restoration, seamless cloning and image stitching. The solution of these algorithms is usually to solve a large sparse linear system: the Poisson equation. However, solving the Poisson equation is a computationally and memory-intensive task that is not suitable for real-time image editing. At the same time, when dealing with large-scale images, such as megapixel images, or even megapixel images, this problem becomes extremely prominent. To solve this problem, researchers have proposed various methods. These methods can be roughly divided into two categories: degenerate space-based solvers and numerical analysis-based solvers. [0003] For the first category, degene...

AI Technical Summary

Problems solved by technology

The idea of ​​the direct method is that the exact solution of the equation can be obtained through finite steps of arithmetic calculations. However, due to the existence of rounding in practice, this type of method can only obtain approximate solutions. The most basic of the direct method is the Gaussian elimination method

The direct solver has poor scalability and is not suitable for solving large linear systems.

The idea of ​​the iterative method is to use a certain limit process to gradually approach the exact solution of the linear equation. The iterative method requires less computer storage units and simple program design, but the iterative solver is not very robust and the convergence speed is slow.

Images