An Image Restoration Method Based On Non-redundant Recombination

Image is an important way for human to obtain information.In today's information society,image appears in every aspect of our life and work.However,in the process of image acquisition and transmission,it is always inevitable to mix with the noise,producing distortion and blur,which reduces the quality of the image.Therefore,the research of image restoration algorithm is very necessary.Especially in recent years,deep learning has become a hot topic,and image restoration as image preprocessing has a great impact on the subsequent work.Although many de-noising algorithms have achieved very good results at low noise level,the de-noising effect at high noise level is often poor.This paper proposes an image restoration method based on non-redundant recombination.By means of down-sampling,the image with high noise level is transformed into four sub-images with low noise level,and each sub-image is denoised.Finally,several sub-images are reconstructed into an image with the same size as the original image,and the reconstructed image is deblurred to get a clear image.The algorithm reduces the noise level of the original image,avoids the noise going down at the high noise level,and improves the denoising effect at the high noise level.It can be seen from the experimental data that the algorithm is effective in both subjective vision and experimental results.This paper is divided into six parts.The first part introduces the background and research status of image restoration.The second part introduces the theoretical basis of image denoising and several classical image denoising algorithms.The third part introduces the basic theory of image deblurring and the classical image deblurring algorithm.The fourth part introduces singular value decomposition and BM3 D algorithm.Finally,an image restoration method based on non-redundant recombination and the experimental results are introduced.