50 fs - 12 ps laser pulses are employed to ablate aluminum, copper, iron, and graphite targets. The ablation-generated momentum is measured with a torsion pendulum. Corresponding time-resolved shadowgraphic measurements show that the ablation process at optimal fluence achieving maximal primarily dominated by photomechanical mechanism. When specific used pulse duration tuned from ps, generated firstly increases then remains almost constant, which could be attributed change of mechanism...

Though deep convolutional neural networks (CNNs) have made great breakthroughs in the field of vision-based gesture recognition, however it is challenging to deploy these high-performance resource-constrained mobile platforms and acquire large numbers labeled samples for training CNNs. Furthermore, there are some application scenarios with only a few or even single one new class so that recognition method based on CNNs cannot achieve satisfactory classification performance. In this paper,...

Due to distribution shift, deep learning based methods for image dehazing suffer from performance degradation when applied real-world hazy images. In this paper, we consider a framework on conditional diffusion models improved generalization real haze. First, find that optimizing the training objective of models, i.e., Gaussian noise vectors, is nontrivial. The spectral bias networks hinders higher frequency modes in vectors being learned and hence impairs reconstruction details. To tackle...

Dynamic characteristics of air plasma generated by focused double collinear femtosecond laser pulses with a time interval 10 ns are experimentally investigated. The emission changes significantly when altering the energy ratio between two pulses. Time-resolved shadowgraphic measurements reveal that small volume transient vacuum is formed inside shock wave produced first pulse, which causes second pulse induced ionization zone to present as separate sections in space. Also recorded strong...

In an optical coherence tomography system, the sidelobes of point-spread function (PSF) introduced from source reduce A-scan imaging resolution and contrast images. A gradual iterative signal subtraction method based on study a point influenced by other points with different distances through PSF is proposed in this paper. Comparing CLEAN algorithm two typical deconvolution methods, processed results demonstrate can effectively least runtime. It also found that it insensitive to noise while...

This paper presents a learning based vessel detection and segmentation method in real-patient ultrasound (US) liver images. We aim at detecting multiple shaped vessels robustly automatically, including with weak ambiguous boundaries. Firstly, candidate regions are detected by data-driven approach. Multi-channel enhancement maps complement performances generated aggregated under Conditional Random Field (CRF) framework. Vessel candidates obtained thresholding the saliency map. Secondly,...

Conventional camera-based hand interaction technique suffered from self-occlusion among fingers, which lowers the detection accuracy of fingertip positions, leading to uncomfortable UI controls. Based on observations, depends postures. We design an framework in is decided response a recognized posture. Using tabletop projection system that has projector and depth sensor, we implement by integrating five touch in-air interactions balance its stability usability.

Recent research has shown Convolutional Neural Networks (CNNs) trained in a fully-supervised fashion achieve promising performance on extreme low-light image denoising task. However, large amount of "noisy-clean" pairs are required to train network, which difficult obtain. In this paper, we propose compact yet effective noise model generate synthetic noisy images for training. Especially, address the severe color distortion problem by identifying novel component, black calibration error, as...

Abstract In modern times, the development of an intelligent system that can automatically detect and recognize poultry diseases is vital for efficient farming reducing human workloads. This paper presents a real-time detector analyze frames captured by monitoring cameras simultaneously chickens identify their healthy status. To overcome challenge appearing small having variant scales in camera frames, we propose integrating scale-aware receptive field enhancement module into YOLOv5 algorithm...

Due to distribution shift, deep learning based methods for image dehazing suffer from performance degradation when applied real-world hazy images. In this paper, we consider a framework on conditional diffusion models improved generalization real haze. First, find that optimizing the training objective of models, i.e., Gaussian noise vectors, is non-trivial. The spectral bias networks hinders higher frequency modes in vectors being learned and hence impairs reconstruction details. To tackle...

A novel noise reduction algorithm is proposed for reducing the and enhancing contrast in 3D Optical Coherence Tomography (OCT) images. First, OCT image divided into two subregions based on local property: background area which additive dominant foreground multiplicative dominant. In background, eliminated by 2D linear filtering combined with frame averaging. foreground, filtering-an extension of filtering. Therefore, denoised reconstructed according to combination foreground. The above...

Physical mechanisms of multiple fluorescence regions and strong scattered light observed in the images micro air plasma generated by double femtosecond laser pulses are investigated using time-resolved shadowgraphic method.

Physical mechanisms of multiple fluorescence regions and strong scattered light observed in the images micro air plasma generated by double femtosecond laser pulses are investigated using time-resolved shadowgraphic method.

Improvement of imaging depth and definition optical coherence tomography (OCT) images are demonstrated with saturated sodium chloride solution as the hyperosmotic agent applied on sample in vivo.