Abstract With the advent of era big data, artificial intelligence has attracted continuous attention from all walks life, and been widely used in medical image analysis, molecular material science, language recognition other fields. As basis intelligence, research results neural network are remarkable. However, due to inherent defect that electrical signal is easily interfered processing speed proportional energy loss, researchers have turned their light, trying build networks field optics,...

Three-dimensional profilometry by sinusoidal fringe projection using phase-shifting algorithms is usually distorted the nonlinear intensity response of commercial video projectors. To overcome this problem, several methods including pulse width modulation (SPWM) were proposed to generate patterns with binary ones defocusing project some certain extent. However, residual errors are nonnegligible for highly accurate measurement fields, especially when level insufficient. In work, we propose...

In medical imaging and astronomical observation, high-resolution (HR) images are urgently desired required. recent years, many researchers have proposed various ways to achieve the goal of image super-resolution (SR), ranging from simple linear interpolation schemes nonlinear complex methods. this paper, we deal with SR reconstruction problem based on theory compressive sensing, which uses a redundant dictionary instead conventional orthogonal basis. We further demonstrate better results...

In this paper, we propose a deep learning method for single infrared image optical noise removal. With fully convolutional neural network, it is able to eliminate the in image. Our architecture consists of two networks: denoising network and conditional discriminator. The takes as input outputs result, while discriminator tries make output look more like target. Actually, only testing phase, feed-forward. Significant quality experiments achieved compared with existing method.

Hyperspectral images (HSIs) with high spatial resolution are challenging to obtain directly due sensor limitations. Deep learning is able provide an end-to-end reconstruction solution from low resolution. Nevertheless, existing deep learning-based methods have two main drawbacks. First, networks self-attention mechanisms often require a trade-off between internal resolution, model performance and complexity, leading the loss of fine-grained, high-resolution features. Second, there visual...

In this paper, we present a simple and effective scene-based nonuniformity correction (NUC) method for infrared focal plane arrays based on interframe registration. This estimates the global translation between two adjacent frames minimizes mean square error properly registered images to make any detectors with same scene produce output value. way, accumulation of registration can be avoided NUC achieved. The advantages proposed algorithm lie in its low computational complexity storage...

Dynamic range reduction and detail enhancement are two important issues for effectively displaying high-dynamic-range images acquired by thermal camera systems. They must be performed in such a way that the high dynamic image signal output from sensors is compressed pleasing manner display on lower monitors without reducing perceptibility of small details. In this paper, new method infrared presented. This maps raw to 8-bit domain based same architecture bilateral filter partitioning...

In this paper, we present a deep learning method for single infrared image stripe noise removal. Our is denoted as convolutional neural network (CNN) that takes the noisy input and outputs clean image. The CNN consists of two components: 1) denoising, substantially removing but losing details, 2) denoising super resolution, completely eliminating residual restore details. exhibits excellent detail preserving performance. Meanwhile it achieves fast speed real-time processing. Experiments...

Infrared thermal imaging is widely used in military and civil fields like security protecting monitoring, reconnaissance, road traffic monitoring. However, most current infrared sensors have limited resolution for their technological limitation. To solve it, we propose a novel end-to-end super-resolution via multiscale spatio-temporal feature fusion network, that can fuse the temporal aligned features of neighboring frames spatial reference frame. In our pyramid module designed as major...

This decade has seen rapid developments in infrared light sensing and imaging, utilizing innovative materials deliberate structural designs. Organic semiconducting are promising the next-generation applications due to their unique properties including solution processability, tunable optoelectronic property, biocompatibility. Spotlight aims provide a snapshot of recent advances organic detection technology, specifically for wavelengths beyond 1000 nm, focusing on performance enhancements new...

Recent studies show that compressed sensing (CS) can recover sparse signal with much fewer measurements than traditional Nyquist theorem. From another point of view, it provides a new idea for super-resolution imaging, like the emergence single pixel camera. However, methods implemented measurement matrix by digital mirror device (DMD) or spatial light modulator, which is serial imaging process and makes method inefficient. In this paper, we propose super resolution system based on parallel...

The skylight polarization pattern, which is a result of the scattering unpolarized sunlight by particles in atmosphere, can be used many insects for navigation. Inspired insects, several navigation sensors have been designed and combined with various heading determination methods recent years. However, up until now, few these studies fully considered influences different meteorological conditions, play key roles accuracy, especially cloudy weather. Therefore, this study makes major...