Image Denoising Based on Dual-tree Complex Wavelet Transform and Convolutional Neural Network

Abstract

In view of the problem that the noise in the image will have a negative impact on the subsequent image processing, this paper proposes a noise reduction algorithm based on the dual-tree complex wavelet transform and convolutional neural network by using the characteristics of the DTCWT and CNN. Firstly, the two-dimensional image is decomposed into two parallel wavelet trees by double tree complex wavelet transform, which forms six high-frequency detail parts and two same low-frequency parts in different directions. Then, it is put into the designed deep convolution neural network for training, Finally, the six high-frequency components and one of the low-frequency components are recombined into the image after noise reduction by using the inverse transform of double tree complex wavelet. The simulation results show that PSNR and MS-SSIM of the image are better than other noise reduction algorithms, and the details of the original image can be restored well while noise reduction. It has practical application value in the case that the collected image has a large noise due to various reasons.