Aiming at the problem of high false alarm rate of infrared small target detection, a detection method of identifying infrared small target by using sample database is proposed. Firstly, according to the imaging characteristics of infrared small target, the background prediction algorithm is used to judge whether it is a suspected target pixel by pixel; Then, according to the prior sample data of the target and background, the random forest algorithm is used to train the classifier, and the classifier is used to make a second decision on the suspected target points to further filter out the false target points; Finally, according to the background sample data marked in the current scene, the false target points in the background are filtered out by using NCC similarity measure, and the final detection result is the infrared small target. Experiments show that this method can effectively reduce the false alarm rate of infrared small target detection, and can be used for infrared small target detection in strong clutter environment.

针对光电载荷对隔振性能的需求，提出一种采用菱形连杆机构作为负刚度组件，具有高静、低动刚度特点的非线性隔振器（简称菱形HSLDS隔振器）。采用静力学分析方法，建立了隔振器数学模型，研究了刚度参数设定以及非线性调节方法；利用谐波平衡法（HBM）求解动力学方程，分析了各参数对隔振性能的影响关系；采用动力学仿真软件ADAMS及实物样机对理论模型与结论进行了验证。测试结果表明：菱形HSLDS隔振器具有较方便的参数调整能力，零位刚度及刚度非线性可通过拉簧参数与连杆参数进行设定、优化，隔振的刚度非线性优化程度受主隔振器阻尼以及零位刚度参数影响。相比于传统线性隔振器，菱形HSLDS具有显著的非线性隔振优势，可较好地满足光电载荷隔振需求。

(Visual Simultaneous Location and Mapping) VSLAM is the key technology of computer vision and equipment intelligence. Firstly, this paper introduces the main algorithms and historical evolution of VSLAM, and summarizes the existing algorithms from the front-end and back-end classification, advantages and disadvantages. Subsequently, new research directions emerged in recent years are introduced, focusing on the combination of depth, semantics and multi-robots. Finally, the existing problems are summarized, and the future direction of development is prospected.

作为主要财富的集结区、战争潜力的潜藏区、国家安全重要的战略地带,城市已成为现代战争的主要战场.通过分析传统机载光电侦察系统在城市反恐中的局限性,阐述了多用户多视窗机载光电侦察系统在城市应用中的前景,并通过对多视窗多用户机载光电侦察设备工作原理的分析,特别是对其中的多用户多视窗技术原理的分析,提出了一种可行的、多用户多视窗光电侦察系统研制方案,为类似产品的研制和应用提供了参考.同时结合当前的人工智能技术,分析了图像自动识别技术在其中应用的可行性,以便在海量信息中筛选出威胁目标并进行实时跟踪,大大提高了侦察监视的效率.

Aiming at the requirements of lightweight,miniaturization and high performance of UAV photoelectric payload,a photoelectric payload controller with multi-task management and servo control function is designed,which is based on dual-heterogeneous OMAPL138 DSP+ARM processor.The 8-channel serial port is extended with the EMIF bus interface.the dynamic dual core debugging method is used which combines GDB and CCS,and the inter-core communication is realized by message queue and circular buffer.The design com-pletes the efficient and accurate control of the photoelectric payload.On one hand,the design improves the volume and power consump-tion of the controller,on the other hand,it improves the speed and reliability of the communication.It has certain engineering application value.

Summarized the mainly algorithms of terrain three-dimensional visualization in virtual reality, for the large-scale terrain data organization in three-dimensional virtual war zone simulation, analyzed the characteristic of change of scene, according to motion regulation and idea of layer and block, presented the organize principle of external static data and terrain blocks index regulation, for the large system pressure in the process of scene update, synthesize available visibility culling methods, for the scene change with large-scale data, the system could operate in a high frame rate.