Ring Artifact Supression Based Super Resolution Reconstruction

Image super-resolution reconstruction(SR)is a digital image processing technique that uses a single or multiple low-resolution(LR)images to reconstruct high-resolution(HR)images.In recent years,it has been widely used in image processing fields.However,SR images are often affected by ringing artifacts,which makes the reconstructed images appear bright and dark ripples.The reasons for ringing artifacts are usually due to the truncation of the image boundary information or the inappropriate selection of the inverse Convolution model,which have seriously affected the quality of the reconstructed image.In view of the above reasons,this paper researches the SR reconstruction algorithm based on ringing suppression.the main results are as follows:1.The research progress of SR reconstruction algorithms are discussed in this paper.Various classical SR algorithms are summarized while the problems existing in the traditional SR algorithms are discussed and the causes of ringing atifacts are analyzed.2.Aiming at the ringing artifacts caused by inaccurate estimation of PSF in motion blurred images,a ringing suppression algorithm based on PSF correction is proposed,which mainly employ a new PSF estimation to replace the PSF parameter estimation in the traditional methods.The new PSF estimation is applied to the improved image continuation and windowing methods to preprocess the image,and then applied to the adaptive Wiener filter framework image reconstruction.Experimental results show that the new algorithm has a significant effect on the boundary ringing and edge ringing.3.Aiming at the problem of ringing caused by SR reconstruction of Poisson noise images,an adaptive TV regularized IBP model suitable for Poisson noise pollution is proposed.The algorithm mainly proposed an adaptive TV model for suppressing the ringing artifact of Poisson noise images,employed an error projection constraint that suppresses Gaussian white noise as a recursive term for this model and substituted into the improved IBP iterative method to reconstruct the high resolution image.Experimental results show that this method can effectively inhibit the ringing artifacts caused by Poisson noise pollution.4.Aiming at the problem that the result HR images of the reconstruction-based algorithims is still not clear enough,a single image SR reconstruction algorithm based on dual model regularization sparse representation is proposed.The algorithm mainly involves sparse representation and constructed the sub-dictionary and ARMA model by sample learning,which are adaptively selected as the regularization terms.The sparse representation and the regularization terms of the image are integrated into the MAP framework for optimization.Experiments show that this method is less affected by ringing artifacts than many reconstruction-based methods,and the reconstruction quality has been greatly improved.5.Finally,the different SR algorithms proposed in this paper are summarized and compared,and the advantages and disadvantages of each algorithm are analyzed.