Research On Super-resolution Reconstruction Method For Image Texture Feature

Super-resolution image reconstruction is an important technology in the field of image processing, and its core idea is using a series of low-resolution images with complementary information to reconstruct one or more images of high resolution images. High resolution images can reflect more details of the scene. Therefore, in the video surveillance, military reconnaissance, remote sensing detection, medical diagnostics and other fields, the super-resolution reconstruction technology has a wide range of applications.In the process of super resolution reconstruction, the texture of the reconstructed image is often blurred, and how to effectively suppress the noise while maintaining the texture feature of the image is the technical difficulty of the reconstruction. To solve this problem, this paper focuses on the method of super resolution reconstruction of image texture features. This paper starts with the research on the image texture feature, and then establishes the function relation between the reconstruction parameters and the image texture features, so as to get the texture adaptive reconstruction model, and finally get the clear texture reconstruction effect.Firstly, the texture feature extraction method is systematically studied in this paper. Secondly, texture features are extracted by the appropriate texture extraction method, and the reconstructed parameters are constructed according to the texture feature. Finally, we do the image super resolution reconstruction. The main work is as follows:1. Texture extraction methods were studied to analyze the advantages and disadvantages of each texture extraction method, and the possibility of each method for super-resolution image reconstruction. Then selected the gray co-occurrence matrix method(GLCM) extract texture features, and established a function of regularization parameters and local image texture features that regularization parameter with the local image texture features adaptive adjustment. The experimental results show that compared with the BTV method, this method can make a better reconstruction effect of image edge and texture details, and also can effectively suppress noise. However, the texture feature extraction based on GLCM method is lack of texture direction information, and its feature extraction time is long, which cannot meet the real-time requirements.2. This paper presents a fast method to extract image texture features, and accordingly build weighting coefficients and regularization parameter model. Experimental results show that this method can not only get high quality image, but also significantly improve the speed of reconstruction, and the reconstruction time is shorter, so that the image can be reconstructed quickly and well.