Topic
Noise reduction
About: Noise reduction is a research topic. Over the lifetime, 25121 publications have been published within this topic receiving 300815 citations. The topic is also known as: denoising & noise removal.
Papers published on a yearly basis
Papers
More filters
TL;DR: A new noise filtering method that combines several filtering strategies in order to increase the accuracy of the classification algorithms used after the filtering process and introduces a noisy score to control the filtering sensitivity.
69 citations
12 Oct 2020
TL;DR: An autoencoder-like neural network is presented, aiming to capture intrinsic structures in point clouds and significantly outperforms state-of-the-art denoising methods under both synthetic noise and real world noise.
Abstract: 3D point clouds are often perturbed by noise due to the inherent limitation of acquisition equipments, which obstructs downstream tasks such as surface reconstruction, rendering and so on. Previous works mostly infer the displacement of noisy points from the underlying surface, which however are not designated to recover the surface explicitly and may lead to sub-optimal denoising results. To this end, we propose to learn the underlying manifold of a noisy point cloud from differentiably subsampled points with trivial noise perturbation and their embedded neighborhood feature, aiming to capture intrinsic structures in point clouds. Specifically, we present an autoencoder-like neural network. The encoder learns both local and non-local feature representations of each point, and then samples points with low noise via an adaptive differentiable pooling operation. Afterwards, the decoder infers the underlying manifold by transforming each sampled point along with the embedded feature of its neighborhood to a local surface centered around the point. By resampling on the reconstructed manifold, we obtain a denoised point cloud. Further, we design an unsupervised training loss, so that our network can be trained in either an unsupervised or supervised fashion. Experiments show that our method significantly outperforms state-of-the-art denoising methods under both synthetic noise and real world noise. The code and data are available at https://github.com/luost26/DMRDenoise
69 citations
TL;DR: In this article, a denoising method for ultrasound medical images, whose quality is degraded by the peculiar phenomenon of speckle noise, is presented, which consists in Gaussian filtering of proper wavelet coefficients of the image, corresponding to vertical and diagonal details.
69 citations
TL;DR: In this paper, the authors developed a methodology for reducing the amplitude of vertical noise sources by 1-2 orders of magnitude, revealing many events that could not be distinguished before noise reduction, and correcting for any signal distortion caused by the noise removal.
Abstract: Through an array of ocean‐bottom seismometers, the Cascadia Initiative is investigating the structure of the Juan de Fuca and Gorda plates. Because the instruments are on the seafloor, they are subject to substantial noise from water waves and bottom currents, especially at long periods. If the seismometer is slightly tilted, some of the high bottom current noise on the horizontals leaks onto the vertical record. Another major type of noise, compliance noise, is created when pressure variations associated with infragravity waves physically deflect the seabed. Extending the work of Crawford and Webb (2000), we developed a methodology for reducing the amplitude of vertical noise sources by 1–2 orders of magnitude, revealing many events that could not be distinguished before noise reduction, and for correcting for any signal distortion caused by the noise removal. We use the horizontal records to predict and remove the tilt noise from the verticals, and we use the pressure records to predict and remove the compliance noise from the first year of Cascadia Initiative data. After determining the degree and direction of tilt for each day at each station, we assessed the success of different instrument designs at minimizing tilt noise; external shielding of the seismometer sensor package is effective at reducing bottom current noise. Tilt at individual instruments can vary continuously throughout a several month deployment and should be determined at least daily for optimal noise removal. By understanding and reducing these noise sources, we hope to open the Cascadia dataset to a wider range of applications.
69 citations
15 Oct 2019
TL;DR: A novel progressive Retinex framework is presented, in which illumination and noise of low-light image are perceived in a mutually reinforced manner, leading to noise reduction low- light enhancement results.
Abstract: Contrast enhancement and noise removal are coupled problems for low-light image enhancement. The existing Retinex based methods do not take the coupling relation into consideration, resulting in under or over-smoothing of the enhanced images. To address this issue, this paper presents a novel progressive Retinex framework, in which illumination and noise of low-light image are perceived in a mutually reinforced manner, leading to noise reduction low-light enhancement results. Specifically, two fully pointwise convolutional neural networks are devised to model the statistical regularities of ambient light and image noise respectively, and to leverage them as constraints to facilitate the mutual learning process. The proposed method not only suppresses the interference caused by the ambiguity between tiny textures and image noises, but also greatly improves the computational efficiency. Moreover, to solve the problem of insufficient training data, we propose an image synthesis strategy based on camera imaging model, which generates color images corrupted by illumination-dependent noises. Experimental results on both synthetic and real low-light images demonstrate the superiority of our proposed approaches against the State-Of-The-Art (SOTA) low-light enhancement methods.
69 citations