Image Interpolation Algorithm Based On A Triangular Grid

Image plays a very important role in daily life and scientific research. It is an important source of information for people. Due to the limitations of the device, many times people can not get high resolution images that they want. Image magnification algorithm is to solve this problem. Image interpolation algorithm is a method of reconstructing a high resolution image using the discrete data of the low resolution image. It plays an important role in the field of image processing, computer vision, medicine and so on.This thesis firstly introduces the research background and basic knowledge of image interpolation algorithm, and illustrates the importance of the image interpolation algorithm. Then the thesis introduces the linear interpolation algorithm, adaptive interpolation algorithm and interpolation algorithm based on edge information, and points out the advantages and disadvantages of different algorithms.The image is obtained by sampling the original curved surface by means of the digital equipment. Image interpolation algorithms reconstruct the original image from the low resolution image in order to achieve the image zooming operation. Since the original surface is very complex, it is difficult to be accurately expressed. At present, the traditional image interpolation algorithm is to construct quadrilateral surface patches on the quadrilateral mesh, and then use all the surface patches to construct the fitting surface of the original surface.In order to overcome the shortcomings of traditional image interpolation algorithms, this thesis proposes a new image interpolation algorithm based on the triangular grid. Different from the traditional image interpolation algorithms, this method constructs the polynomial surface to approximate the original surface based on the triangular grid. This method divides the quadrilateral mesh into the triangular mesh along the direction of edges in the image. Because the triangular mesh can divide the original surface more carefully, it is better for the fitting surface to approximate the original surface. At the same time, it is better to construct the fitting surface on the triangular mesh to keep the edges of the image. Due to the human visual characteristics, image edge quality plays a key role in the visual effect of the image. The visual effect of high resolution images obtained by the new method is greatly improved. Finally, this method is compared with the existing image interpolation algorithms. The experimental results show that the fitting surface constructed on the triangular mesh is better able to approach the original surface. The magnified image obtained by this method has a great improvement in visual effect and objective evaluation criteria.In the end, this thesis summarizes the shortcomings of this algorithm, and points out the direction for further research in the future.