Research And Realization Of Super Resolution Restoration Algorithm

In recent years, the super-resolution image restoration project has been more and more widely. With the corresponding research work, the satellite, video, medical science and other fields have been harvested a certain number and relatively mature research results. For example, the transformation of high-definition digital TV signal without distortion, the significantly increased of CT and Nuclear Magnetic Resonance (NMR) image quality. Can be seen the super resolution figure everywhere. We can say, the super-resolution image restoration has entered into our life.The noise in the image acquisition, transmission process, has been an inevitable factor, and plays a role can’t be ignored in the process of images quality degradation. The influence of the image quality has improved, at the same time, also interfere the process of the signal processing. The traditional Total Variation (TV) method is an effective method for image denoising, but also bring the staircase effect. Based on the traditional Total Variation (TV) method, we used the image space and pixel gradient methods, then changed some improvement, make the edge preserving performance of traditional algorithm be kept. At the same time,to reduce noise in the flat areas, improves the inherent problems in the process of image denoising.Then, the article improved algorithm based on the image denoising, we have combined the prior knowledge and Kullback-Leibler distance. Try to put forward an improved super resolution image restoration algorithm. Use the Kullback-Leibler algorithm to reduce the differences between the distance of the prior model, make the process of finding the approximation to estimate more quickly and effectively. Then estimated the super resolution image. Experiments show that, the experimental effect of this new method is more outstanding then kinds of priori knowledge when they used alone.