Poisson Noise Reduction with Non-local PCA

Salmon, Joseph; Harmany, Zachary T.; Deledalle, Charles-Alban; Willett, Rebecca

doi:10.1007/s10851-013-0435-6

Cited by 267 publications

(210 citation statements)

References 28 publications

Supporting

Mentioning

202

Contrasting

Unclassified

Order By: Relevance

“…′′ 492 pixel). We smoothed these data cubes with the non-local PCA method of Salmon et al (2014). This method combines elements of dictionary learning and sparse patch-based representation of images (or spectral data cubes) for photon-limited data.…”

Section: Expansion Of the Pwnmentioning

confidence: 99%

See 1 more Smart Citation

Expansion and Brightness Changes in the Pulsar-wind Nebula in the Composite Supernova Remnant Kes 75

Reynolds

Borkowski

Gwynne

2018

ApJ

View full text Add to dashboard Cite

We report new Chandra X-ray observations of the shell supernova remnant (SNR) Kes 75 (G29.7−0.3) containing a pulsar and pulsar-wind nebula (PWN). Expansion of the PWN is apparent across the four epochs, 2000, 2006, 2009, and 2016. We find an expansion rate between 2000 and 2016 of the NW edge of the PWN of 0.249% ± 0.023% yr −1 , for an expansion age R/(dR/dt) of 400 ± 40 years and an expansion velocity of about 1000 km s −1 . We suggest that the PWN is expanding into an asymmetric nickel bubble in a conventional Type IIP supernova. Some acceleration of the PWN expansion is likely, giving a true age of 480 ± 50 years. The pulsar's birth luminosity was larger than the current value by a factor of 3 -8, while the initial period was within a factor of 2 of its current value. We confirm directly that Kes 75 contains the youngest known PWN, and hence youngest known pulsar. The pulsar PSR J1846−0258 has a spindown-inferred magnetic field of 5 × 10 13 G; in 2006 it emitted five magnetar-like short X-ray bursts, but its spindown luminosity has not changed significantly. However, the flux of the PWN has decreased by about 10% between 2009 and 2016, almost entirely in the northern half. A bright knot has declined by 30% since 2006. During this time, the photon indices of the power-law models did not change. This flux change is too rapid to be due to normal PWN evolution in one-zone models.

show abstract

Section: Expansion Of the Pwnmentioning

confidence: 99%

“…We use the smoothed profile shown in Figure 2 to model the halo contribution to the background at Epoch II. Since it is Salmon et al (2014). Background has been subtracted as described in the text.…”

Section: Expansion Of the Pwnmentioning

confidence: 99%

Expansion and Brightness Changes in the Pulsar-wind Nebula in the Composite Supernova Remnant Kes 75

Reynolds

Borkowski

Gwynne

2018

ApJ

View full text Add to dashboard Cite

show abstract

“…Under this assumption, summing of the Bernoulli random variables can be thought of as performing a “binning” of the pixels. Binning is a common technique in restoring images from Poisson noise, especially when the signal-to-noise ratio is low [23,25,26]. Binning can also be applied together with transform-denoise, e.g., in [27], to achieve improved results.…”

Section: Non-iterative Image Reconstructionmentioning

confidence: 99%

“…Image reconstruction is a critical component of QIS, for without such an algorithm, we will not be able to form images. However, unlike classical Poisson image recovery problems where solutions are abundant [22,23,24,25,26,27], the one-bit quantization of QIS makes the problem uniquely challenging, and there is a limited number of existing methods [28,29,30,31]. Another challenge we have to overcome is the complexity of the algorithm, which has to be low enough that we can put them on cameras to minimize power consumption, memory consumption and runtime.…”

Section: Introductionmentioning

confidence: 99%

Images from Bits: Non-Iterative Image Reconstruction for Quanta Image Sensors

Chan

Elgendy

Wang

2016

Sensors

View full text Add to dashboard Cite

A quanta image sensor (QIS) is a class of single-photon imaging devices that measure light intensity using oversampled binary observations. Because of the stochastic nature of the photon arrivals, data acquired by QIS is a massive stream of random binary bits. The goal of image reconstruction is to recover the underlying image from these bits. In this paper, we present a non-iterative image reconstruction algorithm for QIS. Unlike existing reconstruction methods that formulate the problem from an optimization perspective, the new algorithm directly recovers the images through a pair of nonlinear transformations and an off-the-shelf image denoising algorithm. By skipping the usual optimization procedure, we achieve orders of magnitude improvement in speed and even better image reconstruction quality. We validate the new algorithm on synthetic datasets, as well as real videos collected by one-bit single-photon avalanche diode (SPAD) cameras.

show abstract

“…However, the xray signal is very weak, so EDS spectrum image data sets are often dominated by Poisson noise. We have applied non-local principle component analysis (NLPCA) [3] to reduce the level of Poisson noise in a model atomic-resolution EDS spectrum image. The processed data has better contrast and lower noise compared to normal methods, enabling more clear visualization of fine spatial features.…”

mentioning

confidence: 99%

Combining Non-Rigid Registration with Non-Local Principle Component Analysis for Atomic Resolution EDS Mapping

Zhang

Yankovich

et al. 2016

Microsc Microanal

Self Cite

View full text Add to dashboard Cite

Energy-dispersive x-ray Spectroscopy (EDS) spectrum imaging in the STEM can be used to create compositionally-sensitive images of materials at atomic resolution [1,2]. However, the xray signal is very weak, so EDS spectrum image data sets are often dominated by Poisson noise. We have applied non-local principle component analysis (NLPCA) [3] to reduce the level of Poisson noise in a model atomic-resolution EDS spectrum image. The processed data has better contrast and lower noise compared to normal methods, enabling more clear visualization of fine spatial features.EDS spectrum images with simultaneous HAADF images were acquired from Ca-stabilized NdTiO3 perovskite with Ca and Nd on the A sites. 100 spectrum image frames were acquired with 256×256 spatial pixels and 2048 energy channels. We applied non-rigid registration (NRR) [4] to the simultaneous HAADF images, and then implemented the shifts to each energy channel of each spectrum image. The resulting registered spectrum images were summed within seven characteristic x-ray lines and result in seven integrated elemental maps.The NLPCA algorithm applies a non-local method for denoising hyperspectral data. It first performs patchization on the original data cube to get a collection of small noisy patches and divides the patches into clusters using k-means clustering. Poisson principle component analysis (PCA) is applied to each cluster to reduce Poisson noise. The final denoised image is reconstructed from the denoised clusters.NLPCA method was applied to the seven integrated band elemental maps to reduce Poisson noise. The final denoised images show atomic structure in all bands except the O Kα band, as shown in Figure 1. The NLPCA results clearly indicate that Ca atoms occupy Nd sites, which is an observation that is not visible before NLPCA but does agree with the known structure.The reliability of the NLPCA algorithm and the best denoising parameters were studied with phantom EDS spectrum image data based on the experimental data but without noise. Three different Ca distributions were considered in the phantom data: a uniform distribution, Ca nanoparticles, and a random distribution. Poisson noise was added to the synthetic data, then denoised using NLPCA. The peak signal-noise ratio (PSNR) was used to measure the denoised image quality and was used to determine the most efficient set of NLPCA parameters. The results, shown in Figure 2, demonstrate that NLPCA is capable of reconstructing the phantom with Poisson noise corrupted data for all seven bands in three different Ca distributions [5].

show abstract

Poisson Noise Reduction with Non-local PCA

Cited by 267 publications

References 28 publications

Expansion and Brightness Changes in the Pulsar-wind Nebula in the Composite Supernova Remnant Kes 75

Expansion and Brightness Changes in the Pulsar-wind Nebula in the Composite Supernova Remnant Kes 75

Images from Bits: Non-Iterative Image Reconstruction for Quanta Image Sensors

Combining Non-Rigid Registration with Non-Local Principle Component Analysis for Atomic Resolution EDS Mapping

Contact Info

Product

Resources

About