Image Magnification Algorithm Based On Parabolic Interpolation And Wavelet Transformation

With the development of information technology, digital images are used in many fields of our daily lives. As a basic operation of image processing, image magnification has a wide usage. The so-called image magnification is that converting a image from a resolution to high resolution by interpolation. This technology is universally applied in medical, public security, military, meteorology, remote sensing, animation and films.Now, the nearest point interpolation, bilinear interpolation and spline interpolation are more mature algorithms. The nearest point interpolation is simple, easily realized, and has faster processing speed than other methods, but this method will engender obvious serrated edge and box-shaped effect; bilinear interpolation algorithm use linear average weights of four neighboring pixel of source pixel as the target pixel value, it is of a certain edge smoothing function, but it will degenerate the detail of the image, and loses important edge feature; images magnified by cubic spline interpolation and cubic B-spline interpolation algorithms have higher smoothness, but it needs lots of calculation, and often make the magnified image fuzzy. The above methods all progress based on the hypothesis that image's pixel and pixels around has linear relation, but actually, there is mutational nature between the image's textures or between pixels, it is nonlinear. In progress of image magnification, if we adopt the conventional interpolation algorithms to generate new pixel to the pixels which are characteristics of discontinuous gray, it will make the figure and texture of magnified image fuzzy, and reduce the quality of the image.This paper presents a new algorithm to magnify images. This algorithm improves the weighted parabolic interpolation algorithm, and enhances the amplitude of low-frequency field of wavelet transformation. In the end, new algorithm uses the improved weighted parabolic interpolation algorithm combine with discrete wavelet transformation to realize the image magnification. The result of experiment shows that, compared with traditional methods, this new algorithm not only keeps the details of the original image, but also enhances the brightness and definition of the amplificatory images.