Search

The performance of deep learning based image super-resolution (SR) methods depend on how accurately the paired low and high resolution images for training characterize the sampling process of real cameras. Low and high resolution (LR $\sim $ HR) image pairs synthesized by degradation models (e.g., bicubic downsampling) deviate from those in reality; thus the synthetically-trained DCNN SR models work disappointingly when being applied to real-world images. To address this issue, we propose a novel data acquisition process to shoot a large set of LR $\sim $ HR image pairs using real cameras. The images are displayed on an ultra-high quality screen and captured at different resolutions. The resulting LR $\sim $ HR image pairs can be aligned at very high sub-pixel precision by a novel spatial-frequency dual-domain registration method, and hence they provide more appropriate training data for the learning task of super-resolution. Moreover, the captured HR image and the original digital image offer dual references to strengthen supervised learning. Experimental results show that training a super-resolution DCNN by our LR $\sim $ HR dataset achieves higher image quality than training it by other datasets in the literature. Moreover, the proposed screen-capturing data collection process can be automated; it can be carried out for any target camera with ease and low cost, offering a practical way of tailoring the training of a DCNN SR model separately to each of the given cameras. [ABSTRACT FROM AUTHOR]

In this work, two electronic nose (E-nose) prototypes for lung cancer detection via breath analysis are built. The instrumental variation between them is a critical problem, as it impacts on the generalization performance of classification model. The conventional domain adaption (DA) methods are based on the assumption that the data quality of both devices is high enough. However, it is not true in our case because the concentrations of volatile organic compounds (VOCs) biomarkers in breath are very low and the acquisition process of breath samples is prone to many confounding factors. As a countermeasure, a novel and effective two-step sparse unidirectional domain adaptation (SUDA) algorithm is proposed. In the first step, the data quality of each domain is improved, at the meantime, the distribution discrepancy between source and target domain is reduced by finding the common discriminative features via the sparse group lasso algorithm. In the second step, the distribution gap is further narrowed down by adopting the strategies of distribution alignment, local geometric characteristics preserving and label dependence constraint. Experimental results show that the proposed method is not only significantly effective for correcting instrumental variation in the two homemade lung-cancer-detection E-noses, but also demonstrates superior performance on a public E-nose instrumental variation dataset. [ABSTRACT FROM AUTHOR]

We propose a novel asymmetric image compression system of light $\ell _\infty $ -constrained predictive encoding and heavy-duty CNN-based soft decoding. The system achieves superior rate-distortion performances over the best of existing image compression methods, including BPG, WebP, FLIF and recent CNN codecs, in both $\ell _{2}$ and $\ell _\infty $ error metrics, for bit rates near or above the threshold of perceptually transparent reconstruction. These remarkable coding gains are made by deep learning for compression artifact removal. A restoration CNN is designed to map a lossy compressed image to its original. Its unique strength is to enforce a tight error bound on a per pixel basis. As such, no small distinctive structures of the original image can be dropped or distorted, even if they are statistical outliers that are otherwise sacrificed by mainstream CNN restoration methods. [ABSTRACT FROM AUTHOR]

The dynamics of the inward-focusing ring Airy beam in nonlocally defocusing nonlinear media is investigated in detail, and a bottle beam shape can be observed by adjusting the amplitude of the inward-focusing ring Airy beam and the nonlocality of the media. The gradient force of the bottle beam is numerically studied. It is worth mentioning that the bottle beams are formed in the nonlinear regime, which indicates that much higher intensity is allowable for these beams than those in the linear regime. Our approach and results may pave the way to the optical tweezers in the nonlinear regime. [ABSTRACT FROM AUTHOR]