CFPNet: a denoising network for complex frequency band signal processing
Ke Zhang,Miao Long,Jie Chen,Mingzhu Liu,Jingjing Li
DOI: https://doi.org/10.1109/tmm.2022.3233398
IF: 7.3
2023-01-01
IEEE Transactions on Multimedia
Abstract:The recent development of deep learning has brought breakthroughs in image denoising. However, the recovery of image detail, especially high-frequency weak information, still needs to be improved. Firstly, the noise mainly concentrates on the high-frequency signal, and the high-frequency signal is easy to be disturbed, which makes it difficult to recover; Secondly, in the process of image denoising with deep learning, feature extraction of model is used to smooth the noise for image restoration, resulting in a poor recovery effect of high-frequency signal. To solve the above problems and improve the overall image denoising performance, we propose a denoising network for complex frequency band signal processing (CFPNet), which contains three insights: 1) the image input node uses a cosine transform to segment the image noise frequency and divides different image features into signals in different frequency bands for targeted noise reduction; 2) targeted noise reduction is carried out for different frequency band signals via a fine-grained scheme; 3) different frequency band signals are fused and high-frequency signals are enhanced to improve the recovery of detailed signals. The experimental results show that the proposed CFPNet can achieve state-of-the-art performance on both real-world datasets and Gaussian noise fitting datasets.
computer science, information systems,telecommunications, software engineering