Reducing boundary artifacts in image deconvolution

Liu, Renting; Jia, Jiaya

doi:10.1109/icip.2008.4711802

Cited by 32 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[28][29][30] Most of these issues become less pronounced when using a confocal microscopy system, which is also quite simpler in its hardware compared to a STED microscope. 31 Using the presented deep learning-based approach, the diffraction induced resolution gap between a STED image and a confocal microscope image can be closed, achieving super-resolution microscopy using relatively simpler and more cost-effective imaging systems, also reducing photo-toxicity and photo-bleaching.…”

Section: Discussionmentioning

confidence: 99%

Deep learning achieves super-resolution in fluorescence microscopy

Wang

Rivenson

Jin³

et al. 2018

Preprint

View full text Add to dashboard Cite

We present a deep learning-based method for achieving super-resolution in fluorescence microscopy. This data-driven approach does not require any numerical models of the imaging process or the estimation of a point spread function, and is solely based on training a generative adversarial network, which statistically learns to transform low resolution input images into super-resolved ones. Using this method, we super-resolve wide-field images acquired with low numerical aperture objective lenses, matching the resolution that is acquired using high numerical aperture objectives. We also demonstrate that diffraction-limited confocal microscopy images can be transformed by the same framework into super-resolved fluorescence images, matching the image resolution acquired with a stimulated emission depletion (STED) microscope. The deep network rapidly outputs these super-resolution images, without any iterations or parameter search, and even works for types of samples that it was not trained for.Computational super-resolution microscopy techniques in general make use of a priori knowledge about the sample and/or the image formation process to enhance the resolution of an acquired image. At the heart of the existing super-resolution methods 1-3 , numerical models are utilized to simulate the imaging process, including, for example, an estimation of the point spread function (PSF) of the imaging system, its spatial sampling rate and/or sensor-specific noise patterns. Fluorescence imaging process is in general more challenging to model and take into account e.g., spatially-varying optical aberrations, the chemical environment of the labeled sample and the optical properties of the specific mounting media and the fluorophores that are used 4-7 . This image modeling related challenge, in turn, leads to formulation of forward models with different simplifying assumptions. In general, more accurate models yield higher quality results, often with a trade-off of exhaustive parameter search and computational cost.Here we present a deep learning-based framework to achieve super-resolution in fluorescence microscopy without the need for making any assumptions on or precise modeling of the image formation process. Instead, we train a deep neural network using a Generative Adversarial Network (GAN) 8 model to transform an acquired low-resolution image into a high-resolution one. Once the deep network is trained (see the Methods section), it remains fixed and can be used to rapidly output batches of high resolution images, in e.g., 0.4 sec for an image size of 1024×1024 pixels using a single Graphics Processing Unit (GPU). The network inference is noniterative and does not require a manual parameter search to optimize its algorithmic performance.The deep network can also be generalized to different types of samples that were not part of the training process.We demonstrate the success of this deep learning-based approach by super-resolving the raw images captured by a widefield fluorescence microscope and a confocal microscope. In the...

show abstract

Section: Discussionmentioning

confidence: 99%

Deep learning achieves super-resolution in fluorescence microscopy

Wang

Rivenson

Jin³

et al. 2018

Preprint

View full text Add to dashboard Cite

show abstract

“…The content of each tiled block is computed by solving a regularized least square problem which imposes smooth transitions within the block itself and with the neighboring blocks' boundaries as well. 23 Since solving the optimization problem can be computationally heavy when large images are being processed, we approximate the content of the tiled blocks by a simple linear interpolation between the given boundary values, followed by adaptively blurring incrementally toward the center of the tiled blocks to impose smoothness. We found that these simple preprocessing steps are sufficient to eliminate the boundary artifacts with a light computation load.…”

Section: Tackling the Boundary Artifactsmentioning

confidence: 99%

Single-image full-focus reconstruction using depth-based deconvolution

et al. 2016

View full text Add to dashboard Cite

CitationSingle-image full-focus reconstruction using depth-based deconvolution 2016, 56 (4):041302 Optical Engineering , "Single-image full-focus reconstruction using depth-based deconvolution," Opt.Abstract. In contrast with traditional extended depth-of-field approaches, we propose a depth-based deconvolution technique that realizes the depth-variant nature of the point spread function of an ordinary fixed-focus camera. The developed technique brings a single blurred image to focus at different depth planes which can be stitched together based on a depth map to output a full-focus image. Strategies to suppress the deconvolution's ringing artifacts are implemented on three levels: block tiling to eliminate boundary artifacts, reference maps to reduce ringing initiated by sharp edges, and depth-based masking to mitigate artifacts raised by neighboring depth-transition surfaces. The performance is validated numerically for planar and multidepth objects.

show abstract

“…The main SSD errors of our results concentrate at the image boundaries, which can be attributed to the boundary value problem inhere in the deconvolution using the fast Fourier transform (FFT). These boundary artifacts can be significantly reduced by Donatelli et al [5], Calvetti et al [32] and Liu et al [33].…”

Section: Objective Experimentsmentioning

confidence: 99%

Multi-scale blind motion deblurring using local minimum

et al. 2009

View full text Add to dashboard Cite

Blind deconvolution, a chronic inverse problem, is the recovery of the latent sharp image from a blurred one when the blur kernel is unknown. Recent algorithms based on the MAP approach encounter failures since the global minimum of the negative MAP scores really favors the blurry image. The goal of this paper is to demonstrate that the sharp image can be obtained from the local minimum by using the MAP approach. We first propose a cross-scale constraint to make the sharp image correspond to a good local minimum. Then the cross-scale initialization, iterative likelihood update and the iterative residual deconvolution are adopted to trap the MAP approach in the desired local minimum. These techniques result in our cross-scale blind deconvolution approach which constrains the solution from coarse to fine. We test our approach on the standard dataset and many other challenging images. The experimental results suggest that our approach outperforms all existing alternatives.

show abstract

Reducing boundary artifacts in image deconvolution

Cited by 32 publications

References 6 publications

Deep learning achieves super-resolution in fluorescence microscopy

Deep learning achieves super-resolution in fluorescence microscopy

Single-image full-focus reconstruction using depth-based deconvolution

Multi-scale blind motion deblurring using local minimum

Contact Info

Product

Resources

About