Since dust particles in the air scatter and absorb light, images captured in sand-dust weather mostly show low contrast, color deviation and blurriness, seriously affecting the reliability of visual tasks. Currently, pixel-level enhancement and prior-based methods are used to restore sand-dust images. However, these methods cannot accurately extract semantic information from the images due to the loss of information and the complexity of the scene depth, which may lead to color distortion and blurred textures of the restored image. We thus presents a two-stage restoration method based on style transformation and unsupervised sand-dust image restoration network (USDR-Net). In the first stage, the grayscale distribution compensation (GDC) method is used to transform the style of the sand-dust image. After transformation, color shift is eliminated and potential information is restored in the balanced image. In the second stage, USDR-Net firstly employs the dark channel prior and the transmission map enhancement network (TME-Network) to generate and refine the transmission map of the balanced image to improve the accuracy of scene depth. Then, it reconstructs a clear image with actual color and high contrast via adversarial learning with unpaired sand-dust and clear images. Extensive experimental results show that our method outperforms state-of-the-art algorithms based on both qualitative and quantitative evaluations. The mean average precision for the target inspection datasets has increased from 16.79% to 68.82%.

To compare the detection performance of near-, mid-, and far-infrared laser fuzes in clouds, we built a laser fuze detection model in a cloud environment based on the Monte Carlo method, and simulated the echoes of near- (0.86 μm), mid- (4.6 μm), and far-infrared (10 μm) pulsed laser fuzes in clouds under two scenarios: the cloud backscattering echo in the absence of a target in clouds and the target echo in the presence of a target in clouds. The echoes of laser fuzes at different wave bands under two scenarios were compared, and the results show that the mid- and far-infrared laser fuzes have comparable detection performance in clouds, while presenting respective advantages and disadvantages compared to the fuze in the near-infrared wave band. With presentation of the detection performance comparison results between the near-, mid-, and far-infrared laser fuzes in clouds, this paper can provide guidance and reference for the application of mid- and far-infrared lasers in the laser fuze field.

For a pulsed laser transmitted in the raised dust from battlefield vehicle movement, the premise of obtaining its transmission properties and anti-jamming method is to acquire the spatial distribution rules of dust mass concentration. In this study, the emission mechanism for vehicle-raised dust in the battlefield environment is analyzed, as well as various influencing factors of the dust emission velocity. A finite volume method (FVM)-based model for simulating the spatial dispersion of dust raised by a certain moving tank is built, and the effects of mass flow rate, traveling speed, wind speed and wind direction on the spatial distribution of dust mass concentration are explored in the battlefield environment. The results show that the raised dust is plume-shaped within the computational domain, and the spatial mass concentration of plume is positively correlated with the mass flow rate. Besides, it is necessary to determine the impact of tank travelling speed on the spatial mass concentration of dust based on the relationship between the travelling speed and the dust mass flow rate. Although front wind reduces the spatial mass concentration of dust plumes, the reduction rate decelerates when the wind speed exceeds 4 m/s. Crosswind leads to the deflection and decreased mass concentration of dust plumes. Under varying wind directions, the variation of crosswind component exerts a more pronounced influence on the spatial mass concentration of dust than that of the front wind component. This study provides a technical support for the further study of near-ground laser weapon system against vehicle dust interference in the battle field.

In land battlefield environments, aerosol particles can cause laser beams to undergo attenuation, thus deteriorating the operational performance of military laser devices. The particle size distribution (PSD) and extinction coefficient are key optical properties for assessing the attenuation characteristics of laser beams caused by aerosol particles. In this study, we employed the laser diffraction method to measure the PSDs of graphite smoke screen, copper powder smoke screen, iron powder smoke screen, ground dust, and soil explosion dust. We evaluated the goodness of fit of six common unimodal PSD functions and a bimodal lognormal PSD function employed for fitting these aerosol particles using the root mean square error (RMSE) and adjusted R², and selected the optimal PSD function to evaluate their extinction coefficients in the laser wavelength range of 0.249~12 μm. The results showed that smoke screens, ground dust, and soil explosion dust exhibited particle size ranges of 0.7~50 µm, 1~400 µm, and 1.7~800 μm, respectively. The lognormal distribution had the best goodness of fit for fitting the PSDs of these aerosol particles in the six unimodal PSD functions, followed by the gamma and Rosin–Rammler distributions. For the bimodal aerosol particles with a lower span, the bimodal lognormal PSD functions exhibited the best goodness of fit. The graphite smoke screen exhibited the highest extinction coefficient, followed by the copper and iron powder smoke screens. In contrast, the ground dust and soil explosion dust exhibited the lowest extinction coefficients, reaching their minimum values at a wavelength of approximately 8.2 μm. This study provides a basis for analyzing and improving the detection and recognition performance of lasers in land battlefield environments.

CNN-based crowd counting methods have achieved great progress in recent years. However, most of these CNN-based crowd counting methods do not make full use of contextual information, which contains high-level semantic features and low-level detail features from different receptive fields of CNN. But rich contextual information is important to solve the scale variation problem of crowd counting. So the precision of previous CNN-based crowd counting methods is decreased. To solve this problem, we propose an adaptive attention fusion mechanism (AAFM). AAFM can use multi-scale features from different receptive fields of CNN effectively. It integrates the convolution network for feature learning and the attention mechanism for multi-scale features fusion. We apply the first 13 convolution layers of VGG-16 as the encoder module to extract the base features. Then, the base features are fed into the decoder module. The decoder module mainly contains Density Regression Branch (DRB) and Feature Fusion Branch (FFB). DRB uses multiple convolution layers for feature learning and multi-scale feature extraction. FFB uses attention modules for modeling multi-scale features and element-wise multiply for features fusion. Therefore, AAFM can obtain rich contextual information into the encoder-decoder framework for generating high-quality crowd density maps and accurate counting. We perform experiments on ShanghaiTech, UCF-CC-50, and UCF-QNRF datasets, and AAFM achieves promising results.

Network traffic demands grow explosively from multimedia, point-to-point, and E-Science applications. In order to meet such tremendous growth, optical networks has been introduced for exploring higher bandwidth and better QoS guarantee. Problems arise when integrating the whole new optical infrastructure with current IP backbone, including controlling data path and network management. In this paper, we discuss the design issues of alarm management in hybrid networks containing DWDM, SDH and IP technologies. For instance, when there is a fiber cut in DWDM layer, tens of alarms would appear concurrently indicating network breakdowns in DWDM, SDH and IP layer separately, which will confuse network engineers for identifying fault location. Therefore, we propose a novel correlation method among these layers according to a pre-defined correlation diagram. Several test scenarios are experimented in out lab in order to illustrate the correlation diagram. Our contribution would benefit network carriers maintaining hybrid networks for root-cause finding among alarms cross different network layers.