A5063129382- Image Enhancement Techniques
- Advanced Image Processing Techniques
- Advanced Image Fusion Techniques
- GaN-based semiconductor devices and materials
- Advanced Vision and Imaging
- Semiconductor Quantum Structures and Devices
- Semiconductor Lasers and Optical Devices
- Color Science and Applications
- Semiconductor materials and devices
- Ga2O3 and related materials
- Photonic and Optical Devices
- Image Processing Techniques and Applications
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Machine Learning and Data Classification
- Optical Coherence Tomography Applications
University of California, Santa Barbara
2022
Dalian University of Technology
2019-2022
Low-light image enhancement plays very important roles in low-level vision areas. Recent works have built a great deal of deep learning models to address this task. However, these approaches mostly rely on significant architecture engineering and suffer from high computational burden. In paper, we propose new method, named Retinex-inspired Unrolling with Architecture Search (RUAS), construct lightweight yet effective network for low-light images real-world scenario. Specifically, building...
Enhancing the quality of low-light (LOL) images plays a very important role in many image processing and multimedia applications. In recent years, variety deep learning techniques have been developed to address this challenging task. A typical framework is simultaneously estimate illumination reflectance, but they disregard scene-level contextual information encapsulated feature spaces, causing unfavorable outcomes, e.g., details loss, color unsaturation, artifacts. To these issues, we...
Achieving high quantum efficiency in long-wavelength LEDs has posed a significant challenge to the solid-state lighting and display industries. In this article, we use V-defect engineering as technique achieve higher efficiencies red InGaN on (111) Si through lateral injection. We investigate effects of superlattice structure distribution, electroluminescence properties, external efficiency. Increasing relative thickness InGaN/GaN total correlate with reduction active region defects...
Distributed feedback laser diodes (DFBs) serve as simple, compact, narrow-band light sources supporting a wide range of photonic applications. Typical linewidths are on the order sub-MHz for free-running III-V DFBs at infrared wavelengths, but short-wavelength GaN-based considerably worse or unreported. Here, we present InGaN DFB operating 443 nm with an intrinsic linewidth 685 kHz continuous wave output power 40 mW. This performance is achieved using first-order embedded hydrogen...
Semantic segmentation is of great value to autonomous driving and many robotic applications, while it highly depends on costly time-consuming pixel-level annotation. To make full use unlabeled data, this work proposes a deep tri-training framework (dubbed DTT) utilize labeled along with data for training in semi-supervised manner. Concretely, the DTT framework, three networks are initialized same structure but different parameters. The optimized circularly, where one network trained each...
Most existing monocular depth estimation approaches are su- pervised, but enough quantities of ground truth data required during training. To cope with this, recent techniques deal the task in an unsupervised man- ner, i.e., replacing use easily obtained stereo images for Based on we propose a nov- el learning architecture, which integrates dual attention mechanism into framework and designs depth- aware loss better estimation. Specifically, to en- hance ability feature representations,...
The under-exposure and low-light environments are common to degrade the image-quality with invisible information. To ameliorate this case, a copious of image enhancement methods developed. However, these existing works hard handle extremely conditions noises, even well-known network-based methods. address issue, we develop Principle-inspired Multi-scale Aggregation Network (PMA-Net) simultaneously achieve exposure noises removal. Specifically, establish pioneering principle-inspired...
Enhancing the quality of low-light images plays a very important role in many image processing and multimedia applications. In recent years, variety deep learning techniques have been developed to address this challenging task. A typical framework is simultaneously estimate illumination reflectance, but they disregard scene-level contextual information encapsulated feature spaces, causing unfavorable outcomes, e.g., details loss, color unsaturation, artifacts, so on. To these issues, we...
Abstract We have successfully demonstrated InGaN/GaN edge-emitting laser diodes (EELDs) on a fully coalesced epitaxial lateral overgrown film from c-plane GaN substrate. achieve high aspect ratio, low defect density wing region covering 75% –88% of the substrates’ surface, which is amongst largest reported area in literature. The devices at exhibit threshold current 3.63 kA/cm² 408 nm, and an improved performance compared to confirmed. Based this work, it promising that performance,...
Low-light image enhancement plays very important roles in low-level vision field. Recent works have built a large variety of deep learning models to address this task. However, these approaches mostly rely on significant architecture engineering and suffer from high computational burden. In paper, we propose new method, named Retinex-inspired Unrolling with Architecture Search (RUAS), construct lightweight yet effective network for low-light images real-world scenario. Specifically, building...
Low-light image enhancement has traditionally been tackled by training a heuristically designed neural network architecture. Despite the success of these approaches, heuristic design pattern inherently not only hinders further optimization architectures, but also limits factors that designer can take into consideration. As result, methods are difficult to achieve balance between enhancing performance and hardware related performance. In this paper, we equip basic algorithm with architecture...