Restoring images distorted by atmospheric turbulence is a ubiquitous problem in long-range imaging applications. While existing deep-learning-based methods have demonstrated promising results in specific testing condi...
详细信息
Restoring images distorted by atmospheric turbulence is a ubiquitous problem in long-range imaging applications. While existing deep-learning-based methods have demonstrated promising results in specific testing conditions, they suffer from three limitations: (1) lack of generalization capability from synthetic training data to real turbulence data;(2) failure to scale, hence causing memory and speed challenges when extending the idea to a large number of frames;(3) lack of a fast and accurate simulator to generate data for training neural networks. In this paper, we introduce the turbulence mitigation transformer (TMT) that explicitly addresses these issues. TMT brings three contributions: Firstly, TMT explicitly uses turbulence physics by decoupling the turbulence degradation and introducing a multi-scale loss for removing distortion, thus improving effectiveness. Secondly, TMT presents a new attention module along the temporal axis to extract extra features efficiently, thus improving memory and speed. Thirdly, TMT introduces a new simulator based on the Fourier sampler, temporal correlation, and flexible kernel size, thus improving our capability to synthesize better training data. TMT outperforms state-of-the-art video restoration models, especially in generalizing from synthetic to real turbulence data.
Single-molecule localization methods play a vital role in a localization-based super-resolution fluorescence microscopy. However, it is difficult for conventional localization schemes based on the Gaussian fitting to ...
详细信息
Single-molecule localization methods play a vital role in a localization-based super-resolution fluorescence microscopy. However, it is difficult for conventional localization schemes based on the Gaussian fitting to locate overlapped high-density fluorescent emitters. Currently, in the spatial domain, the compressive-sensing-based algorithm (CSSTORM) can localize high-emitter-density images. However, the computational cost of this approach is extremely high, which limits its practical application. Here, we propose an alternative frequency-domain compressed sensing (FD-CS) technique for fast super-resolution imaging. Unlike the CSSTORM method, which is a measurement matrix based on the point spread function, a Fourier dictionary designed in the frequency domain and orthogonal matching pursuit is used to reliably recover the original signal. The simulation and experimental results prove that the FD-CS is 1000 times faster than CSSTORM with CVX and ten times faster than that with L1-Homotopy with almost the same localization accuracy and recall rate. This drastic reduction in computational time should allow the compressed sensing approach to be routinely applied to a super-resolution image analysis.
The alignment of the sub-apertures is a major challenge for future segmented telescopes and telescope arrays. We show here that a focal plane wave-front sensor using only two images can fully and efficiently align a m...
详细信息
ISBN:
(数字)9781510619500
ISBN:
(纸本)9781510619500
The alignment of the sub-apertures is a major challenge for future segmented telescopes and telescope arrays. We show here that a focal plane wave-front sensor using only two images can fully and efficiently align a multiple aperture system, both for the alignment (large amplitude tip/tilt aberrations correction) and phasing (piston and small amplitude tip/tilt aberrations correction) modes. We derive a new algorithm for the alignment of the sub-apertures : ELASTICS. We quantify the novel algorithm performance by numerical simulations. We show that the residues are within the capture range of the fine algorithms. We also study the performance of LAPD, a recent real-time algorithm for the phasing of the sub-apertures. The closed-loop alignment of a 6 sub-aperture mirror provides experimental demonstration for both algorithms.
Super-resolution (SR) is a class of techniques that enhance the resolution of an imaging system by combining complimentary information from several images to produce high resolution images of a subject. Fast non-itera...
详细信息
ISBN:
(数字)9783319118543
ISBN:
(纸本)9783319118536
Super-resolution (SR) is a class of techniques that enhance the resolution of an imaging system by combining complimentary information from several images to produce high resolution images of a subject. Fast non-iterative and iterative algorithms are described in this article. The metrics to compare the images are investigated also. In conclusion shows the comparative results of these methods. Test results showed good practical applicability of the developed algorithms.
We demonstrate an efficient algorithm for the temporal and spatial based calculation of the laser speckle contrast analysis (LASCA) for the imaging of blood flow that reduces the numerical complexity of necessary calc...
详细信息
ISBN:
(纸本)9780819496508
We demonstrate an efficient algorithm for the temporal and spatial based calculation of the laser speckle contrast analysis (LASCA) for the imaging of blood flow that reduces the numerical complexity of necessary calculations, facilitates a multi-core implementation of the speckle analysis and enables an independence of temporal or spatial resolution and SNR. The new algorithm was evaluated for both spatial and temporal based analysis of speckle patterns with different image sizes and incorporated pixels as sequential and multi-core code. The improvement is about a factor of 5 and can be increased to about a factor of 15 for multi-core computers. This allows an online-analysis of larger speckle images or at a higher frame rate.
It is known that in the field of image synthesis, multi-frame image processing can produce better results than just single-image enhancement techniques. Such a technique uses similarity assessment to select the images...
详细信息
ISBN:
(纸本)9781424450466
It is known that in the field of image synthesis, multi-frame image processing can produce better results than just single-image enhancement techniques. Such a technique uses similarity assessment to select the images for synthesis. In this paper, we propose a new image similarity assessment index by using complex wavelet. It is found that the proposed index is robust to small rotations and translations as well as large intensity and contrast changes. Therefore, it can be widely applied in the field of image synthesis, especially high dynamic range imaging.
暂无评论