Super Resolution Image Reconstruction With Multi Dictionary Via Edge Perserving

Super resolution image reconstruction is a technique for low resolution image processing. It aims to recover the high frequency information of the low resolution image, so that it will become a high resolution image with rich content and good vision. This technology has aroused strong reaction from domestic and foreign scholars. After a long time exploration and research, the technology has formed several branches, and the reconstruction method based on learning has become a hot research field in recent years because of its relatively good reconstruction effect. This paper is to reconstruct an image using the method of this branch; the main contents are as follows:Firstly, the reconstruction effect is related to the relevance of the training sample, and the traditional method is to reconstruct all the images in a single dictionary, which is lack of pertinence. To solve this problem, this paper presents an improved method of image super-resolution reconstruction based on multi dictionary, take image guided classification to establish multi dictionary and take KPCA method to improve the reconstruction speed. The experimental results show that the proposed method has higher efficiency and faster speed compared with the traditional method.Secondly, as the traditional classification method classifies edge and texture image blocks depend on the threshold, and it needs amounts of computation. This paper presents an improved image block classification method, it though canny operator to extract the edge, and then carry on the guidance of the training image blocks by using edge block. By comparing the traditional method with the improved classification method, the results show that the improved classification method can achieve better reconstruction results compared with the traditional classification method.Finally, for the reconstruction of the traditional algorithm of the image edge will appear jagged and other artificial traces, this paper aims to improve the reconstruction of the edge effect. Though researching the reconstruction method based on gradient profile prior find that it has a good edge preserving property, so the method of multi dictionary super-resolution reconstruction via edge preserving is proposed. The correlation of image blocks classification is improved by the gradient prior reconstruction. It can also improve the edge effect of the reconstructed image. Experimental results show that the proposed method is not only superior to the traditional method but also the multi dictionary method, and it has a good effect both on the visual perception and objective evaluation.