Filters








9,701 Hits in 3.6 sec

Denoising as well as the best of any two denoisers

Erik Ordentlich
2013 2013 IEEE International Symposium on Information Theory  
Finally, we show that for the binary symmetric channel, combining the loss estimation with a randomization step leads to a solution to the stated problem under no restrictions on the given denoisers.  ...  We show that under certain restrictions on the two given denoisers the problem can be solved using a straightforward application of a known loss estimation paradigm.  ...  The lemma implies that for any BSC the estimate (19) of the randomized denoiser conditional expected loss concentrates for all clean sequences and all underlying denoisers, including those in which the  ... 
doi:10.1109/isit.2013.6620452 dblp:conf/isit/Ordentlich13 fatcat:kq2dodxqfza7batqarqharh6be

Denoising as well as the best of any two denoisers [article]

Erik Ordentlich
2020 arXiv   pre-print
Finally, we show that for the binary symmetric channel, combining the loss estimation with a randomization step leads to a solution to the stated problem under no restrictions on the given denoisers.  ...  We show that under certain restrictions on the two given denoisers the problem can be solved using a straightforward application of a known loss estimation paradigm.  ...  Following (5), the estimated unnormalized loss of denoiser 1, on the other hand, is N 0 and of denoiser 2, 0.  ... 
arXiv:2007.05905v1 fatcat:be5jorvsajcrtngyg457i5ez2i

Neural Universal Discrete Denoiser [article]

Taesup Moon, Seonwoo Min, Byunghan Lee, Sungroh Yoon
2016 arXiv   pre-print
Unlike other approaches that utilize supervised learning for denoising, we do not require any additional training data.  ...  We present a new framework of applying deep neural networks (DNN) to devise a universal discrete denoiser.  ...  Second, we plan to give theoretical analyses on the concentration (12) and justify the derived k selection rule.  ... 
arXiv:1605.07779v2 fatcat:7xq5fyeytffn3kc6s2tckybmzq

A Denoising Loss Bound for Neural Network based Universal Discrete Denoisers [article]

Taesup Moon
2018 arXiv   pre-print
The key tool we develop is the concentration of the unbiased estimated loss on the true denoising loss, which is shown to hold uniformly for all bounded network parameters and all underlying clean sequences  ...  by minimizing the empirical estimated loss.  ...  The key result for obtaining such bound was to show the uniform concentration of the average estimated loss on the average true denoising loss in Theorem 2.  ... 
arXiv:1709.03657v2 fatcat:7tlm3akcdvfk7n2tqns2md4iw4

Self-supervised Dynamic CT Perfusion Image Denoising with Deep Neural Networks [article]

Dufan Wu, Hui Ren, Quanzheng Li
2020 arXiv   pre-print
In this paper, we proposed a self-supervised deep learning method for CTP denoising, which did not require any high-dose reference images for training.  ...  The method was validated on both simulation and a public real dataset. The proposed method achieved improved image quality compared to conventional denoising methods.  ...  The network took one time-frame image x(t) and one early frame image x(t 0 ) as input and directly output the denoised time-concentration image c d (t).  ... 
arXiv:2005.09766v1 fatcat:d4puiinfkfdctjg7drdt35l5ty

Wavelet denoising for voxel-based compartmental analysis of peripheral benzodiazepine receptors with 18F-FEDAA1106

Miho Shidahara, Yoko Ikoma, Chie Seki, Yota Fujimura, Mika Naganawa, Hiroshi Ito, Tetsuya Suhara, Iwao Kanno, Yuichi Kimura
2007 European Journal of Nuclear Medicine and Molecular Imaging  
For clinical data, there were visual improvements in the signal-to-noise ratio for estimated BP images.  ...  Purpose We evaluated the noise reduction capability of wavelet denoising for estimated binding potential (BP) images (k 3 /k 4 ) of the peripheral benzodiazepine receptor using 18 F-FEDAA1106 and nonlinear  ...  Noise propagation for estimated physiological parameters depends on the particular kinetic model and on the estimation procedure [15] .  ... 
doi:10.1007/s00259-007-0623-y pmid:18026949 fatcat:a5u6eykttnhjte7v5cwvnqn5dq

Point Cloud Denoising via Momentum Ascent in Gradient Fields [article]

Yaping Zhao, Haitian Zheng, Zhongrui Wang, Jiebo Luo, Edmund Y. Lam
2022 arXiv   pre-print
To achieve point cloud denoising, traditional methods heavily rely on geometric priors, and most learning-based approaches suffer from outliers and loss of details.  ...  Recently, the gradient-based method was proposed to estimate the gradient fields from the noisy point clouds using neural networks, and refine the position of each point according to the estimated gradient  ...  Loss Function. In Section 3.3, we describe the loss term for training the gradient estimation network. 3. Momentum Ascent Denoising Algorithm.  ... 
arXiv:2202.10094v2 fatcat:lkw3yi6wqfehdmkltcw4xstd6y

Reduction of Compton Background Noise for X-ray Fluorescence Computed Tomography with Deep Learning

Peng Feng, Yan Luo, Ruge Zhao, Pan Huang, Yonghui Li, Peng He, Bin Tang, Xiansheng Zhao
2022 Photonics  
In the training process, the L1 loss function is used for its good convergence.  ...  In this paper, a noise2noise denoising algorithm based on the UNet deep learning network is proposed. The network can use noise image learning to convert the noise image into a clean image.  ...  point estimation problem separately for each input sample.  ... 
doi:10.3390/photonics9020108 fatcat:r476lur4tva2xcqquyccybkgvy

Twice-Universal Denoising

Erik Ordentlich, Krishnamurthy Viswanathan, Marcelo J. Weinberger
2013 IEEE Transactions on Information Theory  
We consider a class of denoisers that apply one of a number of constituent denoisers based on minimizing an estimated denoising loss and establish a formal relationship between errors in the estimated  ...  We consider a class of denoisers that apply one of a number of constituent denoisers based on minimizing an estimated denoising loss and establish a formal relationship between errors in the estimated  ...  k = o(n/ log n), the estimated loss of a given k-th order sliding window denoiser concentrates around the true loss.  ... 
doi:10.1109/tit.2012.2216503 fatcat:oosqcuxrnfboxoi6kgjdqfohbi

Contrastive Blind Denoising Autoencoder for Real-Time Denoising of Industrial IoT Sensor Data [article]

Saúl Langarica, Felipe Núñez
2021 arXiv   pre-print
In this work, a purely data-driven self-supervised learning-based approach based on a blind denoising autoencoder is proposed for real time denoising of industrial sensor data.  ...  Blind denoising is achieved by using a noise contrastive estimation (NCE) regularization on the latent space of the autoencoder, which not only helps to denoise but also induces a meaningful and smooth  ...  Fig. 6 . 6 CBDAE results when blind denoising the output solids concentration, one of the key variables of the thickener for control purposes.  ... 
arXiv:2004.06806v2 fatcat:o6p7sivp4vaerl5pkq7lkl374m

Synergy Between Semantic Segmentation and Image Denoising via Alternate Boosting [article]

Shunxin Xu, Ke Sun, Dong Liu, Zhiwei Xiong, Zheng-Jun Zha
2021 arXiv   pre-print
The proposed network is composed of multiple segmentation and denoising blocks (SDBs), each of which estimates semantic map then uses the map to regularize denoising.  ...  We observe that not only denoising helps combat the drop of segmentation accuracy due to noise, but also pixel-wise semantic information boosts the capability of denoising.  ...  After that, some works [1, 5] proposed different networks for denoising, which concentrated on the structure of networks.  ... 
arXiv:2102.12095v1 fatcat:fbnz55d24ncydod7yaclmrm7sm

DFT-Based Channel Estimation and Noise Variance Estimation Techniques for Single-Carrier FDMA

Gillian Huang, Andrew Nix, Simon Armour
2010 2010 IEEE 72nd Vehicular Technology Conference - Fall  
Denoise Channel Estimator • The filter coefficients for the denoise channel estimator are: • At low SNR, the noise is large compared to the smeared channel energy in the energy smearing region, and this  ...  As most of channel energy will be concentrated in a few taps, a time domain filter (one tap per sample) can be designed to reduce the estimation error.  ...  Conclusions • The commonly used LS channel estimator gives about 3dB performance loss compared to the ideal-LMMSE channel estimator.  ... 
doi:10.1109/vetecf.2010.5594158 dblp:conf/vtc/HuangNA10 fatcat:2jdiildukfcbrn53o462v2m6lu

Identifying Recurring Patterns with Deep Neural Networks for Natural Image Denoising [article]

Zhihao Xia, Ayan Chakrabarti
2019 arXiv   pre-print
One recourse is to rely on "internal" image statistics, by searching for similar patterns within the input image itself.  ...  The denoising algorithm then aggregates matched coefficients to obtain an initial estimate of the clean image.  ...  Thus, we adopt a pre-training strategy using a loss defined on pairs of patches at a time, using a simplified loss for denoising patch i by averaging it with patch j as: L ij = g s g i − r g i 2 + m g  ... 
arXiv:1806.05229v3 fatcat:w67maizedjh73kpceum6tnjnuu

PET kinetic analysis: wavelet denoising of dynamic PET data with application to parametric imaging

Miho Shidahara, Yoko Ikoma, Jeff Kershaw, Yuichi Kimura, Mika Naganawa, Hiroshi Watabe
2007 Annals of Nuclear Medicine  
This review describes wavelet denoising of dynamic PET images for improving image quality in estimated parametric images.  ...  Wavelet denoising provides signifi cantly improved quality directly to dynamic PET images and indirectly to estimated parametric images.  ...  An improvement in SNR was achieved in the formation of parametric images of the concentration of receptors available for binding (B′ max ) without a signifi cant loss in spatial resolution.  ... 
doi:10.1007/s12149-007-0044-9 pmid:17876550 fatcat:p4kqtp2eczapvczn5xozvafzni

Simultaneous Determination of Metal Ions in Zinc Sulfate Solution Using UV–Vis Spectrometry and SPSE-XGBoost Method

Fei Cheng, Chunhua Yang, Can Zhou, Lijuan Lan, Hongqiu Zhu, Yonggang Li
2020 Sensors  
Therefore, our developed approach can be implemented as a promising mean for real-time and on-line determination of multi-metal ion concentrations in zinc hydrometallurgy.  ...  We developed an accurate and rapid approach based on the singular perturbation spectrum estimator and extreme gradient boosting (SPSE-XGBoost) algorithms to simultaneously determine multi-metal ion concentrations  ...  For example, the denoising spectrum has a positive effect on copper, cobalt, and nickel but negative on zinc; (ii) derivative pretreatment provides more abundant and effective data for spectral prediction  ... 
doi:10.3390/s20174936 pmid:32878223 pmcid:PMC7506957 fatcat:xwza7m5h45gyndi3sl7ygcpbt4
« Previous Showing results 1 — 15 out of 9,701 results