Application of Deep Learning-Based Methods to the Single Image Non-Uniform Blind Motion Deblurring Problem

In this paper, we present deep learning-based blind image deblurring methods for estimating and removing a non-uniform motion blur from a single blurry image. We propose two fully convolutional neural networks (CNN) for solving the problem. The networks are trained end-to-end to reconstruct the latent sharp image directly from the given single blurry image without estimating and making any assumptions on the blur kernel, its uniformity, and noise. We demonstrate the performance of the proposed models and show that our approaches can effectively estimate and remove complex non-uniform motion blur from a single blurry image.

3D Printing of Customized Aspheric Lenses for Imaging

A simple and efficient process for fabricating customized aspheric lenses is reported, in which a stereolithographic 3D printer combined with the meniscus equilibrium post-curing technique is employed. Two kinds of UV-curable resins, DentaClear and HEMA, were used for printing aspheric lenses in our experiments. The printed DentaClear lens featured low surface profile deviation of ~74 μm and showed satisfactory optical imaging resolution of 50.80 lp/mm, i.e., 4.92 μm. The surface roughness of the printed lens with DentaClear was measured to be around 2 nm with AFM. The surface roughness was improved as a result of post-curing, which reduced the ripples on printed lens surfaces. In contrast, the printed HEMA lens exhibited a significant stair-stepping effect with a large surface profile deviation of ~150 μm. The ripples were somewhat apparent even if the printed HEMA lens surface was smoothed by means of post-curing. No sharp image can be obtained with the HEMA lens in the resolution testing. The composition of HEMA resin may be the reason for the relatively poor surface quality and optical properties.

Dual Image Deblurring Using Deep Image Prior

Blind image deblurring, one of the main problems in image restoration, is a challenging, ill-posed problem. Hence, it is important to design a prior to solve it. Recently, deep image prior (DIP) has shown that convolutional neural networks (CNNs) can be a powerful prior for a single natural image. Previous DIP-based deblurring methods exploited CNNs as a prior when solving the blind deburring problem and performed remarkably well. However, these methods do not completely utilize the given multiple blurry images, and have limitations of performance for severely blurred images. This is because their architectures are strictly designed to utilize a single image. In this paper, we propose a method called DualDeblur, which uses dual blurry images to generate a single sharp image. DualDeblur jointly utilizes the complementary information of multiple blurry images to capture image statistics for a single sharp image. Additionally, we propose an adaptive L2_SSIM loss that enhances both pixel accuracy and structural properties. Extensive experiments show the superior performance of our method to previous methods in both qualitative and quantitative evaluations.

Uncertainty Quantification in Image Segmentation Using the Ambrosio–Tortorelli Approximation of the Mumford–Shah Energy

The quantification of uncertainties in image segmentation based on the Mumford–Shah model is studied. The aim is to address the error propagation of noise and other error types in the original image to the restoration result and especially the reconstructed edges (sharp image contrasts). Analytically, we rely on the Ambrosio–Tortorelli approximation and discuss the existence of measurable selections of its solutions as well as sampling-based methods and the limitations of other popular methods. Numerical examples illustrate the theoretical findings.

Deep Learning-Enabled Variational Optimization Method for Image Dehazing in Maritime Intelligent Transportation Systems

Image dehazing has become a fundamental problem of common concern in computer vision-driven maritime intelligent transportation systems (ITS). The purpose of image dehazing is to reconstruct the latent haze-free image from its observed hazy version. It is well known that the accurate estimation of transmission map plays a vital role in image dehazing. In this work, the coarse transmission map is firstly estimated using a robust fusion-based strategy. A unified optimization framework is then proposed to estimate the refined transmission map and latent sharp image simultaneously. The resulting constrained minimization model is solved using a two-step optimization algorithm. To further enhance dehazing performance, the solutions of subproblems obtained in this optimization algorithm are equivalent to deep learning-based image denoising. Due to the powerful representation ability, the proposed method can accurately and robustly estimate the transmission map and latent sharp image. Numerous experiments on both synthetic and realistic datasets have been performed to compare our method with several state-of-the-art dehazing methods. Dehazing results have demonstrated the proposed method’s superior imaging performance in terms of both quantitative and qualitative evaluations. The enhanced imaging quality is beneficial for practical applications in maritime ITS, for example, vessel detection, recognition, and tracking.

Perbaikan Citra Menggunakan Metode Contrast Stretching

Contrasting images that are not good because they are too bright or too dark cannot provide good information. Therefore, a method is needed to improve the image quality, so that the information in the image can be conveyed properly. Contrast stretching is one of the methods for improving image quality. With this method is expected to produce a new image that is better. The purpose of this research is to apply contrast stretching method to an application or software that can be used to improve image quality. Data used in this study in the form of grayscale image data and RGB imagery (true color), with the format . BMP or .JPG, while the application development uses the Matlab programming language.The results of the study, contrast stretching method can be used to repair image that affects bad or poor image quality such as too bright / dark image, less sharp image, blurry, and so on. Contrast stretching method can also be used to improve image enhancement by leveling the histogram that was collected in an area, so that the information contained in the image is more clearly visible compared to the original image.

Relative depolarization of the black hole photon ring in GRMHD models of Sgr A* and M87*

ABSTRACT Using general relativistic magnetohydrodynamic simulations of accreting black holes, we show that a suitable subtraction of the linear polarization per pixel from total intensity images can enhance the photon ring feature. We find that the photon ring is typically a factor of ≃2 less polarized than the rest of the image. This is due to a combination of plasma and general relativistic effects, as well as magnetic turbulence. When there are no other persistently depolarized image features, adding the subtracted residuals over time results in a sharp image of the photon ring. We show that the method works well for sample, viable GRMHD models of Sgr A* and M87*, where measurements of the photon ring properties would provide new measurements of black hole mass and spin, and potentially allow for tests of the ‘no-hair’ theorem of general relativity.

Backprojection Autofocus of Large Ships with Arbitrary Motion for Synthetic Aperture Radar

Radar images of the open sea taken by airborne synthetic aperture radar (SAR) show typically several smeared ships. Due to their non-linear motions on a rough sea, these ships are smeared beyond recognition, such that their images are useless for classification or identification tasks. The ship imaging algorithm presented in this thesis consists of a fast image reconstruction using the fast factorized backprojection algorithm and an extended autofocus algorithm of large moving ships. This thesis analysis the factorization parameters of the fast factorized backprojection algorithm and describes how to choose them nearoptimally in order to reconstruct SAR images with minimal computational costs and without any loss of quality. Furthermore, this thesis shows how to estimate and compensate for the translation, the rotation and the deformation of a large arbitrarily moving ship in order to reconstruct a sharp image of the ship. The proposed autofocus technique generates images in which the ...