Image denoising can be described as the problem of mapping from a noisy image to a noise-free image. The best currently available denoising methods approximate this mapping with cleverly engineered algorithms. In this work we attempt to learn this mapping directly with a plain multi layer perceptron (MLP) applied to image patches. While this has been done before, we will show that by training on large image databases we are able to compete with the current state-of-the-art image denoising methods. Furthermore, our approach is easily adapted to less extensively studied types of noise (by merely exchanging the training data), for which we achieve excellent results as well.

/pdf/image-denoising-can-plain-neural-networks-compete-with-bm3d-1p3aavylzd.pdf

Image denoising: Can plain neural networks compete with BM3D?

Recently, very deep convolutional neural networks (CNNs) have been attracting considerable attention in image restoration. However, as the depth grows, the longterm dependency problem is rarely realized for these very deep models, which results in the prior states/layers having little influence on the subsequent ones. Motivated by the fact that human thoughts have persistency, we propose a very deep persistent memory network (MemNet) that introduces a memory block, consisting of a recursive unit and a gate unit, to explicitly mine persistent memory through an adaptive learning process. The recursive unit learns multi-level representations of the current state under different receptive fields. The representations and the outputs from the previous memory blocks are concatenated and sent to the gate unit, which adaptively controls how much of the previous states should be reserved, and decides how much of the current state should be stored. We apply MemNet to three image restoration tasks, i.e., image denosing, super-resolution and JPEG deblocking. Comprehensive experiments demonstrate the necessity of the MemNet and its unanimous superiority on all three tasks over the state of the arts. Code is available at https://github.com/tyshiwo/MemNet.

MemNet: A Persistent Memory Network for Image Restoration

We present an extension of the BM3D filter to volumetric data. The proposed algorithm, BM4D, implements the grouping and collaborative filtering paradigm, where mutually similar d -dimensional patches are stacked together in a (d+1) -dimensional array and jointly filtered in transform domain. While in BM3D the basic data patches are blocks of pixels, in BM4D we utilize cubes of voxels, which are stacked into a 4-D “group.” The 4-D transform applied on the group simultaneously exploits the local correlation present among voxels in each cube and the nonlocal correlation between the corresponding voxels of different cubes. Thus, the spectrum of the group is highly sparse, leading to very effective separation of signal and noise through coefficient shrinkage. After inverse transformation, we obtain estimates of each grouped cube, which are then adaptively aggregated at their original locations. We evaluate the algorithm on denoising of volumetric data corrupted by Gaussian and Rician noise, as well as on reconstruction of volumetric phantom data with non-zero phase from noisy and incomplete Fourier-domain (k-space) measurements. Experimental results demonstrate the state-of-the-art denoising performance of BM4D, and its effectiveness when exploited as a regularizer in volumetric data reconstruction.

Nonlocal Transform-Domain Filter for Volumetric Data Denoising and Reconstruction

Lacking realistic ground truth data, image denoising techniques are traditionally evaluated on images corrupted by synthesized i.i.d. Gaussian noise. We aim to obviate this unrealistic setting by developing a methodology for benchmarking denoising techniques on real photographs. We capture pairs of images with different ISO values and appropriately adjusted exposure times, where the nearly noise-free low-ISO image serves as reference. To derive the ground truth, careful post-processing is needed. We correct spatial misalignment, cope with inaccuracies in the exposure parameters through a linear intensity transform based on a novel heteroscedastic Tobit regression model, and remove residual low-frequency bias that stems, e.g., from minor illumination changes. We then capture a novel benchmark dataset, the Darmstadt Noise Dataset (DND), with consumer cameras of differing sensor sizes. One interesting finding is that various recent techniques that perform well on synthetic noise are clearly outperformed by BM3D on photographs with real noise. Our benchmark delineates realistic evaluation scenarios that deviate strongly from those commonly used in the scientific literature.

/pdf/benchmarking-denoising-algorithms-with-real-photographs-182c5avqmf.pdf

Benchmarking Denoising Algorithms with Real Photographs

In this paper, we consider a typical image blind denoising problem, which is to remove unknown noise from noisy images. As we all know, discriminative learning based methods, such as DnCNN, can achieve state-of-the-art denoising results, but they are not applicable to this problem due to the lack of paired training data. To tackle the barrier, we propose a novel two-step framework. First, a Generative Adversarial Network (GAN) is trained to estimate the noise distribution over the input noisy images and to generate noise samples. Second, the noise patches sampled from the first step are utilized to construct a paired training dataset, which is used, in turn, to train a deep Convolutional Neural Network (CNN) for denoising. Extensive experiments have been done to demonstrate the superiority of our approach in image blind denoising.

/pdf/image-blind-denoising-with-generative-adversarial-network-3iuvpetrsp.pdf

Image Blind Denoising with Generative Adversarial Network Based Noise Modeling

In order to resolve the problem that the denoising performance has a sharp drop when noise standard deviation reaches 40, proposed to replace the wavelet transform by the DCT. In this comment, we argue that this replacement is unnecessary, and that the problem can be solved by adjusting some numerical parameters. We also present this parameter modification approach here. Experimental results demonstrate that the proposed modification achieves better results in terms of both peak signal-to-noise ratio and subjective visual quality than the original method for strong noise.

Comments on "Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering

In this paper, we develop a new image denoising method based on block-matching and transform-domain filtering The developed method is derived from the current state-of-the-art denoising method (BM3D) We separate the 3D transform in the original method to two steps 1D transform, to further enhance the sparsity for signals whose elements are highly similar and to weaken the sparsity for those signals whose elements are dissimilar Because the 1D filtering is on highly similar elements and the 2D filtering on image blocks are all removed, the image details can be better reserved and fewer artifacts are introduced than original method Experimental results demonstrate that the developed method is competitive and better than some of the current state-of-the-art denoising methods in terms of peak signal-to-noise ratio, structural similarity, and subjective visual quality© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE) Downloading of the abstract is permitted for personal use only

Image denoising by block-matching and 1D filtering

In this paper, we develop a novel transform called semisubsampled wavelet transform (SSWT) and employ it to image watermarking. SSWT consists of two parts, one is nonsubsampled tight frame transform, the other is critically sampled wavelet transform (WT). Embedding watermark into the low-frequency sub-band of SSWT, the imperceptibility and robustness of watermark can be significantly improved comparing with some existing watermarking schemes. Experimental results show that proposed blind watermarking scheme is robust against JPEG compression, Gaussian noise, Wiener filtering and median filtering attacks. For rotation attack, we propose a novel watermarking resynchronization approach, the ideal watermark can be always successfully extracted after resynchronization operation to any angle rotated watermarked image. Experimental results show that the proposed resynchronization approach is considerably effective and feasible.

Semisubsampled Wavelet Transform Based Image Watermarking with Strong Robustness to Rotation Attacks

The paper proposed a novel algorithm for texture classification system. This texture classification system is based on the extracted features on the performance of texture images’ Nonsubsampled Contourlet Transform (NSCT). To decrease the dimension of feature vector, we achieve the mean and standard deviation of NSCT coefficients matrix in different subbands and various directions. To compare the performance of the proposed algorithm, the other two commonly used algorithms including of Wavelet Package Transform and the improved LBP descriptor are used to extract texture features. This paper presents the application of Least Square Support Vector Machine (LS-SVM) classifiers to realize the automatic texture classification. The performed numerical experiments show that our algorithm produces a marked improvement in classification performance, which suggests a significant advance in texture classification.

Deyun Yang

Papers

Comments on "Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering

Image denoising by block-matching and 1D filtering

Semisubsampled Wavelet Transform Based Image Watermarking with Strong Robustness to Rotation Attacks

Texture Classification Using Nonsubsampled Contourlet Transform and LS-SVM

New Algorithm for Texture Classification based on Fisher-Markov Selector