- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Solving Inverse Computational Imaging Problems using Deep Pixel-level Prior
摘要： Signal reconstruction is a challenging aspect of computational imaging as it often involves solving ill-posed inverse problems. Recently, deep feed-forward neural networks have led to state-of-the-art results in solving various inverse imaging problems. However, being task specific, these networks have to be learned for each inverse problem. On the other hand, a more flexible approach would be to learn a deep generative model once and then use it as a signal prior for solving various inverse problems. We show that among the various state of the art deep generative models, autoregressive models are especially suitable for our purpose for the following reasons. First, they explicitly model the pixel level dependencies and hence are capable of reconstructing low-level details such as texture patterns and edges better. Second, they provide an explicit expression for the image prior which can then be used for MAP based inference along with the forward model. Third, they can model long range dependencies in images which make them ideal for handling global multiplexing as encountered in various compressive imaging systems. We demonstrate the efficacy of our proposed approach in solving three computational imaging problems: Single Pixel Camera (SPC), LiSens and FlatCam. For both real and simulated cases, we obtain better reconstructions than the state-of-the-art methods in terms of perceptual and quantitative metrics.
关键词： lensless image reconstruction,MAP inference,Inverse problems,compressive image recovery,autoregressive models,deep generative models
更新于2025-09-10 09:29:36
-
One-bit compressive sensing with time-varying thresholds in synthetic aperture radar imaging
摘要： In this study, the authors introduce a new framework for 1-bit compressed synthetic aperture radar (SAR) imaging by using time-varying thresholding. They show how to recover sparse SAR images from noisy measurements which have been quantised to 1-bit with time-varying thresholds. In the conventional 1-bit compressive sensing (CS) SAR imaging methods, 1-bit quantisation is implemented by comparing the received signal to a zero threshold. This makes the information about the magnitude of the signal to be lost and exact signal recovery becomes impossible. One-bit quantisation with time-varying thresholds allows them to reconstruct the magnitude of the signal more accurately and an explicit unit-norm constraint is no longer required in the proposed optimisation formulation. Using the proposed approach, the authors formulate 1-bit CS SAR imaging reconstruction problem as an unconstrained optimisation problem where the objective function includes an ?2 data-fidelity term and a non-smooth regularisation function. In order to solve this unconstrained optimisation problem, they use variable splitting and the alternating direction method of multipliers based approach which is computationally efficient and easy to implement. The results from experiments with synthetic and real SAR images validate the effectiveness of the proposed method named as BCST-SAR (binary CS with time-varying thresholds in SAR imaging).
关键词： ADMM,time-varying thresholds,sparse signal recovery,1-bit compressive sensing,synthetic aperture radar
更新于2025-09-10 09:29:36
-
[IEEE 2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR) - Philadelphia, PA (2018.8.6-2018.8.8)] 2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR) - Theoretical Limitations of Visual Navigation of Lifesaving VSV using Small UAS
摘要： This paper presents a theoretical analysis of the problem of visual navigation of an unmanned surface vehicle (USV) using visual feedback from a small unmanned aerial system (sUAS) in drowning victim recovery scenario. Manually teleoperated life-saving USVs have been used by responders to assist drowning victims. However, the teleoperation is difficult and requires additional manpower. In this project, the rescue is automated by using a sUAS to navigate the USV to the victims autonomously. This enables the responders to focus on the actual rescue. In this paper, the focus is given to the theory behind the problem. The problem is divided into visual position estimation, visual orientation estimation, and motion planning subproblems. For each subproblem, a theoretical analysis is provided and lower bound limits on how well the subproblem can be solved are determined. This will help to understand the theoretical limitations and to drive further research. Lastly, the previously published system is compared with the lower bound limits to identify the areas of potential improvement.
关键词： unmanned surface vehicle,theoretical analysis,small unmanned aerial system,visual navigation,drowning victim recovery
更新于2025-09-10 09:29:36
-
Recovery of scratch deformation formed on crosslinked polyorganosiloxane films
摘要： As consumers generally desire longer lasting scratch-free coatings, scratch resistance has become one of the most important properties of automotive coatings. Surface scratches that deform a coating but do not break it can recover over time, eventually even becoming invisible to the human eye. This phenomenon arises from elastic-plastic behavior and is called "self-recovery". This paper mainly focuses on the recovery of scratch deformations. Conventional coatings with low glass transition temperature (Tg) have fast recovery characteristics, like rubbers, but cannot attain sufficient hardness. Therefore, it is difficult to simultaneously improve both scratch recovery characteristics and hardness in conventional coatings such as 2 K urethane coatings consisting of acrylic polyols because both arise from the same causes as the mobility of the main chain of the coating. Herein, we explore the remarkable improvement of scratch recovery characteristics without sacrificing hardness by using a base resin with a siloxane main chain instead of a carbon-carbon main chain such as that of an acrylic polyol. We demonstrate that crosslinked polyorganosilsesquioxane coatings (Si-films) achieve both hardness suitable for automobile coatings and improved scratch recovery characteristics superior to those of conventional coatings (C-films). Si-films demonstrated 20% better elastic recovery index than C-films, and the Si-films recovered about twice as fast as the C-films. Scratch deformations on the Si-films almost disappeared after 15 min, even at room temperature, while C-films were needed over 24 days. Thus, coatings based on a siloxane main chain may offer dramatically improved self-recovery.
关键词： Scratch resistance,Coatings,Scratch deformation,Recovery,Polyorganosiloxane,Automotive
更新于2025-09-10 09:29:36
-
Recovery of α-Al2O3 from Ionizing Radiation Dosimetric Sensors
摘要： Corundum, sapphire or α-Al2O3 is an important technological material in many optical and electronic applications such as solid-state lasers, optical windows and, more recently, as a radiation detector. Landauer (Glenwood, IL, USA) accumulated large number of archived and stored Luxel? dosemeters composed of Al2O3:C, which were subjected to a recovery process. Due to the importance of this advanced crystalline material in OSL dosimetry, a recovery process was developed based on the dosemeters calcination and Al2O3:C has been reused in manufacturing of new dosemeters. This paper does not aim to optimize the recovery process, but provides an opportunity to study the involved process parameters and to recover this valuable crystalline material from used dosemeters. To the best of our knowledge no other recovery process involving this dosimetric material was described in scientific literature. Fourier Transform Infrared Spectrometry (FTIR), Thermogravimetry/Differential Thermoanalysis (TG/DTA), Differential Scanning Calorimetry (DSC), X-ray Diffraction (XRD), Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES), Optically Stimulated Luminescence (OSL) and Rietveld Refinement were used to characterize the recovered material and to check for the stability of its structural and dosimetric properties.
关键词： recovery process,optically stimulated luminescence,Al2O3:C
更新于2025-09-10 09:29:36
-
Rationalizing and Controlling the Surface Structure and Electronic Passivation of Cesium Lead Halide Nanocrystals
摘要： Colloidal lead halide perovskite nanocrystals (NCs) have recently emerged as versatile photonic sources. Their processing and luminescent properties are challenged by the lability of their surfaces, i.e. the interface of the NC core and the ligand shell. On the example of CsPbBr3 NCs, we model the nanocrystal surface structure and its effect on the emergence of trap states using density functional theory. We rationalize the typical observation of a degraded luminescence upon aging or the luminescence recovery upon post-synthesis surface treatments. The conclusions are corroborated by the elemental analysis. We then propose a strategy for healing the surface trap states and for improving the colloidal stability by the combined treatment with didodecyldimethyl ammonium bromide and lead bromide and validate this approach experimentally. This simple procedure results in robust colloids, which are both highly pure and exhibit high photoluminescence quantum yields of up to 95-98%, retained even after 3-4 rounds of washing.
关键词： trap states,didodecyldimethyl ammonium bromide,luminescence recovery,photoluminescence quantum yields,lead bromide,CsPbBr3 NCs,density functional theory,Colloidal lead halide perovskite nanocrystals,surface structure
更新于2025-09-09 09:28:46
-
Exact Camera Location Recovery by Least Unsquared Deviations
摘要： We establish exact recovery for the Least Unsquared Deviations (LUD) algorithm of "Ozye\c sil and Singer. More precisely, we show that for sufficiently many cameras with given corrupted pairwise directions, where both camera locations and pairwise directions are generated by a special probabilistic model, the LUD algorithm exactly recovers the camera locations with high probability. A similar exact recovery guarantee for camera locations was established for the ShapeFit algorithm by Hand, Lee, and Voroninski, but with typically less corruption.
关键词： robust estimation,random graph theory,structure from motion,least unsquared deviations,camera location estimation,convex recovery
更新于2025-09-09 09:28:46
-
based colour image restoration
摘要： In this paper, we introduce a total variation l1-l2 regularization scheme with adapting the parameter for image restoration involving blurry and noisy colour images. Numerically, an efficient augmented Lagrangian method associated with alternating minimization method is described to obtain the optimal solution recursively. We provide the convergence analysis for the resulting algorithm. Experimental results show that our proposed model and algorithm have good the signal to noise ratio (SNR) and the improvement in signal to noise ratio (ISNR) values for a motion blur with different kinds of noises.
关键词： convex minimization problem,Augmented Lagrangian,image recovery problems,image restoration,total variation (TV)
更新于2025-09-09 09:28:46
-
Machine Learning for Perovskites' Reap-Rest-Recovery Cycle
摘要： Perovskite photovoltaics are efficient and inexpensive, yet their performance is dynamic. In this Perspective, we examine the effects of H2O, O2, bias, temperature, and illumination on device performance and recovery. First, we discuss pivotal experiments that evaluate perovskites’ ability to go through a reap-rest-recovery (3R) cycle, and how machine learning (ML) can help identify the optimum values for each operating parameter. Second, we analyze perovskite dynamics and degradation, emphasizing the research challenges surrounding this 3R cycle. We then outline experiments that could identify the impact of environmental factors on recovery for different perovskite compositions. Finally, we propose an ML paradigm for maximizing long-term performance and predicting device performance recovery, including a shared-knowledge repository. By reframing perovskites’ optoelectronic transiency within the context of recovery rather than degradation, we highlight a set of research opportunities and the artificial intelligence solutions needed for the commercial adoption of these promising solar cell materials.
关键词： machine learning,perovskite photovoltaics,environmental factors,reap-rest-recovery cycle,device performance
更新于2025-09-09 09:28:46
-
ISAR IMAGING BASED ON L1 L0 NORMS HOMOTOPY 2D BLOCK SPARSE SIGNAL RECOVERY ALGORITHM
摘要： Many traditional sparse signal recovery based ISAR imaging methods did not utilize the block scatterers information of targets. Some block Bayesian learning based ISAR imaging algorithms are computational expensive. In this paper, a 2D block (cid:2)1 (cid:2)0 norms homotopy sparse signal recovery algorithm (the BL1L0 algorithm) is proposed and utilized to form the ISAR image. Compared with Bayesian-based algorithms, this algorithm can obtain ISAR images with similar image quality, but the computation speed is faster. Real data experiments verify the merits of our algorithm.
关键词： homotopy algorithm,ISAR imaging,sparse signal recovery,computational efficiency,block scatterers
更新于2025-09-09 09:28:46