Search

Thermoreflectance imaging technique is a noncontact measurement technology that has attracted much attention in the field of thermal characteristics measurement of semiconductor devices. This technology is often used for steady-state temperature measurement and transient temperature measurement of semiconductor devices, with the advantages of high spatial resolution, high temperature resolution, and high time resolution. However, during the measurement of the device, the thermal expansion of the temperature control table and the sample caused by the temperature rise makes the optical image appear defocused. In order to ensure that the optical image collected by the CCD is not distorted, it is necessary to design an image alignment system for focusing so that the semiconductor device can be measured normally. In this paper, an image alignment system consisting of an imaging module, a high-precision motion control component and an image alignment algorithm is proposed to solve the problem of image defocusing caused by thermal expansion, which can provide references for the optimal design and precise measurement of the thermal reflection imaging system based on CCD.

In sea battlefield search and rescue operation problems, the risk assessment for ships is a vital procedure to get a decision support as an important intelligence to combat search and rescue. In this paper, based on back propagation (BP) neural network, a risk assessment indicator system is proposed first to facilitate the hazard, the vulnerability, and the resilience. Next, by developing two types of network structure, the main neural network for estimating the risk has been modeled. Then, the corresponding learning algorithm is designed. Finally, 69 samples for search and rescue ships are given as the input of the neural network. The performance and regression state of the obtained training results demonstrate that the proposed risk assessment method is valid and effective.

Since first reported publicly in Wuhan, the novel coronavirus pneumonia (COVID-19) has attracted the attention all over the world. As one of the worst-hit places, Italy had registered 1,479,910 tested cases, with 211,938 infected cases, 82,879 recoveries, 29,079 deaths as of 29 June 2020. The Italian authorities have implemented multiple control strategies, such as basic social-distancing measures, early measures (including closing the mall and school), and even nationwide lockdown. Herein, a novel model, named as S2E2IHRD, is presented to reveal the course of epidemic in Italy, where eight groups of the population are considered: S, susceptible cases (unquarantined); Sq, susceptible cases (quarantined); E, exposed cases (infected, undetected); Eq1, exposed cases (infected, quarantined); I, infected cases (confirmed, undetected); H, hospitalized cases; R, recovered cases; D, dead cases. Our model distinguishes the infections based on whether they are quarantined or hospitalized. The essential distinction between quarantined/hospitalized and unquarantined/non-hospitalized cases is that the former are typically isolated who appear to unlikely spread the infection. The implementation of countermeasures has been explained in the model, the short-term simulation results have demonstrated the accuracy of our model by comparing with real data on the COVID-19 epidemic in Italy. We also model various simulation scenarios to illustrate the performance of countermeasures.

Maritime search and rescue is the last line of defense to ensure the safety of life and property at sea. To solve the problem of cooperative planning of regional coverage task for multiple search and rescue equipment in the actual search and rescue process. Firstly, we propose a weighted region model mapped by the Monte Carlo drift prediction model. And then we use three reinforcement learning algorithms (Q-learning, Sarsa, Sarsa (lambda)) for comparative experiments. Finally, three algorithms are evaluated per the coverage and repetition rate, and the experimental results show that Sarsa coverage rate is higher and repetition rate is lower. We have solved the weighted area coverage problem in the real search and rescue process, greatly improving the decision-making efficiency of the system, and making the search with the highest probability, high search coverage and low repetition rate, which has extremely high practical value.

Hand tracking is an important part of Human-Computer Interaction (HCI). Computer vision based hand tracking can free the users and help HCI more natural. Hand area should be cropped from the image before hand tracking. Inspired by the success of deep learning, we use CNN to crop the hand area. We proposed a new method named FACPM to crop hand area with attention mechanism and Feature pyramid networks. We adopt VGG as base network, to improve the speed of the framework, unimportant network is cut off. FPN model is used for different scale features and can perform better on tiny feature. Attention model is used for distributing the weights of different features in FPN. A better cropped image can lead to a more precise 3D hand pose estimation and can be used in hand tracking.

Terrorism is a common threat to mankind, and fighting terrorism is the responsibility of every country. Aiming at suspects prediction of terrorist attacks, five methods based on machine learning are used to sort the recorded data of terrorist attacks quantitatively. Based on Logical Regression, Decision Tree, Gaussian Bayesian Network, AdaBoost and Random Forest fitting model, the terrorist attacks recorded, Global Terrorism Database(GTD), in 2015 and 2016 were used as experimental data. The most likely suspects for each attack and the possibility of these attacks, which may be launched by the five most harmful terrorist organizations, will be sorted out and the ranking of these attacks launched by each terrorist organization in the incident will be shown out. The results show that Decision Tree has the highest accuracy and the precision among the candidates, and Gaussian Bayesian Network is close behind.

The military technology planning is a key driving force for improving combat effectiveness. With regard to technological management support, the research on the architecture has been relatively weak. This paper proposed a model based technological management framework for defense acquisition. By analyzing the core decision elements in the technological management process, we define the domain-specific meta-model of the framework to capture the different needs of stakeholders. And then, six products of the technology view (TV) and a specific development process are designed to organize the concepts and relationships into models to support decision-making from different perspectives. The data consistency of models will be ensured by the meta-model. As a consequence, technical analysis activities can be performed based on the data and models in this technology architecture. This model-based approach ensures the availability, integrity, shareability, and compatibility of the architecture. Ultimately, the technology of shipborne unmanned helicopters (SUH) was explored by the ModelLink to demonstrate the applicability and effectiveness of the architecture.

A multi-perspective describing method for emotional speech is proposed. Three perspectives - cognitive, psychological, physiological ones and seven phonetic evaluation features are involved, with which it is expected to describe speech emotion more completely and to bridge the gap between emotions and acoustic signals. A series of experiments including cognition tests and corpus annotation are implemented. Although preliminarily, the results still reveal the effectiveness of the method.

This paper examined the UNESCO PBL teacher training's effect through the investigation, and what problem still exist in the training, then we found the training effect is good, and this training really effective. So under the project based learning environment, the teacher training pattern can promote high effective teacher training, this can provide reference for the later teacher training.