The Image Zooming Methods Based On Partial Differential Equations

Image zooming (also known as image super-resolution) was a behavior which was changing the size of the given image in order to meet people's visual and application requirements. It was widely used in actuality, such as image transmission, high-definition digital TV, film synthesis and so on. In recent years, the image super-resolution had caused the extensive concern of the domestic and overseas researchers. About the image zooming, there were two main classes of methods, including the methods based on partial differential equations and the interpolation methods. In this thesis, based on reviewing the classical interpolation-enlargement methods and the present main methods, a class of new nonlinear partial differential equation models for image zooming was proposed. The details were as follows,(1) For avoiding the block effects and the edge blurring of the traditional image interpolation zooming methods, a novel nonlinear partial differential equation model to post-processing the interpolation zooming image was proposed, which not only efficiently modified the defects of the interpolation zooming image, but also ensured reaching the real-time effects. The experimental results showed that this method was effective.(2) The model of the section (1) based on the classical interpolation methods and the post-processing with partial differential equations was complicated. Therefore, an easier model of nonlinear partial differential equations was also proposed in this paper, which can be directly used for image zooming processing not only decreased computational complexity, but also ensured the image super-resolution effect. The experimental results showed this method was also effective.