Research On The Adaptive Regularization For Image Super-resolution Reconstruction

Image Super-resolution reconstruction is a processing that a set of lower resolution images with displacement and complementary information into one or multiple higher resolution images. Image Super-resolution reconstruction has been widely applied to many fields, such as remote sensing monitoring, video monitoring, high sensitivity digital television, medicine.This paper conducted a comprehensive research on the theory and algorithm of image Super-resolution reconstruction. In order to improve the quality of image edge and image texture area, this paper completely analyzed and research the adaptive regularization for image super-resolution reconstruction method.In this paper, the main contribution can be summarized as follows:1. A super-resolution algorithm based adaptive weighted total variation was proposed. In tradition bilateral total variation regularization method, to the scale weight is a constant, so the image reconstruction effect is not ideal for retain the image details and denoising. The adaptive kernel function is used to control the shape and bandwidth of image. According to this character an adaptive weighted regularization function and regularization parameter algorithm is propose in this paper. This method can retain the image details and remove the image noise better.2. In order to improve real-time of the image super-resolution reconstruction, using the pyramid LK optical flow method with bilateral total variation regularization for image reconstruction and based on MATLAB and C/C ++ mixed programming and Open CV method to achieve super-resolution video images.