Research On Image Super-resolution Reconstruction Based On Anchored Neighborhood Regression

Image super-resolution reconstruction technology can break through the limitation of imaging hardware and reconstruct low-resolution image into high-resolution image.Therefore,image super-resolution reconstruction technology has been applied in the fields of satellite remote sensing,monitoring security,medical image imaging and so on.In recent years,the image super-resolution reconstruction algorithm based on anchored neighborhood regression has attracted much attention,because of its fast reconstruction speed and high image quality.Therefore,we mainly study the image super-resolution reconstruction algorithm based on anchored neighborhood regression.The main work includes the following two points:(1)In order to further improve the quality of image reconstruction,the iterative backprojection algorithm is used as the preprocessing algorithm of anchored neighborhood regression super-resolution reconstruction algorithm.In order to solve the problem that it is difficult to set the backprojection filter,a method of solving backprojection filter based on linear regression is proposed.Firstly,the global mapping matrix which can represent the relationship between high and low resolution images is solved by linear regression,and the backprojection filter is calculated by the global mapping matrix.The experimental results show that the iterative backprojection algorithm as the preprocessing algorithm of anchored neighborhood regression super-resolution reconstruction algorithm can improve the quality of imagereconstruction.(2)Based on the anchored neighborhood regression algorithm,in order to ensure the quality of the reconstructed image,the feature needs to find the nearest anchor point in 1024 anchor points.Therefore,in order to reduce the computational complexity of the nearest anchor point of feature search,a new hash code is proposed,and the hash code is applied to the training stage and reconstruction stage of the anchoring neighborhood regression super-resolution algorithm.In the training stage,the features with the same coding value are clustered by K-means to get the anchor point set,and the coded value is used as the label of the anchor point set.In the reconstruction phase,by hashing the feature,the anchor point set matching the coded value is quickly found,and the nearest anchor point is found in the K anchor points.The experimental results show that the anchoring neighborhood regression super-resolution algorithm based on hash coding is superior to the original algorithm in reconstruction quality and reconstruction speed.