An Surface Fitting Image Super-resolution Algorithm Based On Triangle Mesh Partition

In the process of collecting and acquiring images,the quality of the collected images will be reduced due to defocus,noise and other factors.With the rapid development of computer technology and image processing technology,people's demand for image quality is increasing day by day.However,due to the environment and the limitation of hardware equipment,the quality of the collected images is far below the expectation.In order to meet people's needs,image super-resolution algorithm is generated.Image magnification algorithm is to improve the resolution of the image through a certain function or model.And at the same time to ensure the image edge and texture structure.It can improve the overall visual effect of the image.It is widely used in medical imaging,virtual reality,military,communications and other fieldsFirstly,This paper introduces the background and related knowledge of image super-resolution algorithm.Then,several common image super-resolution algorithms are discussed.Wc analyze the basic principle and performance of each method and compare their advantages and disadvantages.On this basis,this paper presents an image super-resolution algorithm based on triangular meshes.Different from the traditional image interpolation algorithm which uses quadrilateral mesh for interpolation,this method reconstructs the fitting surface on the triangle mesh to approximate the original scene surface.LBP algorithm and second order difference quotient are used to divide the triangular mesh accurately.And then the quadratic polynomial surface at each pixel is solved.In this process,we introduce the edge angle as the constraint,so that the information at the edge of the constructed surface is more abundant.Then,the weighted average of area coordinates is introduced to obtain triangular surface slices,and the initial super-resolution image is obtained by resampling the fitted surface.Finally,we use a global structure sparse regularization strategy to optimize the initially enlarged image.This strategy can further eliminate artifacts at image edges and texturesIn this paper,we propose a new method to divide the pixel grid,which is the LBP algorithm combined with simple second-order difference.The method is simple and can effectively extract the strong edges and texture details of image.It is helpful to improve the quality of the reconstructed image to construct the fitting surface by dividing the mesh with edge constraint.