Research And Implementation Of Image Super-resolution Reconstruction System

In recently, with the increasing popularity of the Internet, the mobile communications industry develops very rapidly and a variety of multimedia services(high-definition television, video surveillance, digital libraries, video telephony, video conferencing, etc.) continue to emerge. And image and video technology becomes widely used. Objective world’s scenes contain a variety of information, but due to the limits of imaging conditions of hardware devices and image acquisition system, getting all the information of the original scene is impossible, and the limits of the hardware device will lead the lost of some high frequency details. The super-resolution reconstruction technique uses some certain algorithms to overcome the restrictions of imaging system and achieve the purpose of obtaining high-quality images from low-resolution images. So super-resolution reconstruction algorithm has a broad development space.This paper proposes a single image super-resolution reconstruction algorithm, and analyses the existing super-resolution image reconstruction algorithm. The main work as follows:1.Aimed at the super-resolution reconstruction algorithm based on interpolation, this paper describes and analyses the traditional nearest neighbor interpolation algorithm, bilinear interpolation algorithm, bi-cubic difference algorithm and Newton interpolation algorithm. The traditional Newton interpolation algorithm only uses the information of the images in one direction, and the loss of image information in the other direction resulting the reconstructed image’s edge blurring, so the image quality is not high. This paper improves the Newton interpolation algorithm. The algorithm uses the information in the original image pixel from two directions, so it reduces the loss of image-related information, and more fully utilizes the information contained by the image. And it is conducive to improving the quality of interpolated images.2.This paper introduces the principle of the wavelet-based super-resolution reconstruction algorithm. Aiming at the characteristics of the Contourlet wave, such as multi-resolution, multi-scale, multi-directional, this paper proposed a new super-resolution image reconstruction algorithm which based on wavelet Contourlet and improved Newton interpolation algorithm. Experimental results show that the algorithm improves the reconstructed image’s clarity and has better protection of the details of the image. And the edge is more legible and it also improves the visual effect of the image. Based on the results, this paper builds a single super-resolution image reconstruction system.