Fast Image Super-resolution Based On Symmetric Hyper-cone Hash Algorithm

In the fields of medical imaging,aerospace,safety monitoring and others,the quality of the acquired image is poor and the resolution is low because of the hardware limitation of the imaging equipment or the imaging environment.These degraded images are not conducive to the subsequent processing and research.It is very important to improve the resolution and the quality of the image.Image super-resolution reconstruction technology can break through the hardware constraints,rebuild a high-quality,high-resolution image from one or more low-resolution images.At present,the learning-based super-resolution reconstruction technology has been widely concerned by researchers because of the high quality of reconstruction and the advantage of rapid reconstruction.We also study the linear regression model of super-resolution reconstruction technique and propose a new algorithm:Fast Image Super-resolution Based on Symmetric Hyper-cone Hash Algorithm.In this paper,the effectiveness of the algorithm is proved by comparison experiment.The innovation of this algorithm mainly has the following two aspects:1.This paper presents a new vector similarity criterion:Absolute Value of Cosine Similarity(AVCS).For the case where the antipodal points(two points on the sphere are antipodal if they are opposite through the center)exist in the set of eigenvector points,the vector points are similar to each other using AVCS instead of the Euclidean distance judgment.Compared with the Euclidean distance,the computation complexity of AVCS is smaller,and the AVCS calculated the nearest neighbor AVCS value is larger,indicating that the nearest neighborhood is more accurate.2.We proposes a new approximate regressor search algorithm:Symmetric Hyper-cone Hash algorithm(SHcH).The SHcH algorithm uses the AVCS criterion to determine whether the feature vector points belong to a certain symmetric hyper-cone surface.Compared with the traditional spherical hash algorithm,the encoding speed of the eigenvector points is improved and the antipodal points can be placed in the same bucket(The intersection of different symmetrical hyper-cone surfaces is called a bucket),so that the same regressor can applied to the antipodal points.Through the SHcH algorithm,we can dramatically reduce the amount of regressor search by losing very little image performance,thus quickly reconstructing a high-resolution image.