Complex Contourlet Single Image Super-Resolution Algorithm Design

Super-resolution analysis of digital image can be widely used in aerospace,precise navigation and precise attack,crop disease diagnosis,natural disaster prediction,identification and certification,and many other fields.It can be divided into two kind problems depending on the type of the source of the low-resolution image: one is super-resolution algorithm of multiple(or sequential)images,another is super-resolution algorithm of single image.In this paper,we study the super-resolution algorithm of single image.Firstly,we discuss the basic concept of digital image super-resolution in this paper,and divide low resolution the images which needs to interpolate into smooth part and non-smooth part.In view of the two parts,it has been studied of super-resolution algorithm of single digital image.The first part discusses the spatial interpolation method for smooth and continuous function model,mainly includes: the nearest neighbor method,bilinear interpolation method,double three interpolation method,cubic spline method and two-dimensional direct interpolation method.The experimental results show that the two-dimensional direct interpolation method was better than the front four kinds of interpolation algorithm,and it can get the optimal super-resolution results under the same complexity condition.However,the nearest neighbor method shows the worst performance in all cases and appears the worst Mosaic phenomenon.Double three interpolation method and cubic spline method is superior to the nearest neighbor method and bilinear interpolation method.In the second part,we use complex contour wavelet transform for interpolation algorithm.So far,there are two main types of complex contour wavelet transformation,namely fixed redundancy complex contour wavelet transform and variable redundancy complex contour wavelet transform.The former fixed transform is dual tree complex wavelet transform,it can be thought of as real part and imaginary part of the basic outline of the wavelet transform.According to the different circumstance,the later variable transform can select different redundancy which is based on the mapped architecture.This article selects the latter as a tool for this transformation because of its higher flexibility and restructuring characteristics.This article expounds the realization method of complex contourlet transform,and discusses the basic properties of this kind of transformation,especially focus on thetransform of the same level,on the basis of the transform;we design a good performance of single image interpolation algorithm.The algorithm fully applies the invariant properties of the complex contour move wavelet transform,the low-resolution need illustrations as a non-smooth part,we interpolate the transform domain within the outline of the wave field,and then reverse transformation,based on this,we will have a fusion for this algorithm and two-dimensional direct interpolation algorithm,we obtained the good experimental results at last.The experimental results show that it is better to realize single image super-resolution reconstruction of the presented algorithm,and it has good performance both in the objective parameters(such as signal to noise ratio of peak)and subjective effect.