一种基于特征匹配的目标识别跟踪方法

针对复杂背景下的动态目标跟踪问题,提出了一种基于边缘检测,综合多图像特征与伺服机构位置信息进行匹配的目标识别跟踪方法。利用SUSAN算法检测边缘,提取出单帧图像中的可疑目标,依次选用灰度、目标几何、伺服机构位置信息和边界不变矩信息匹配,完成目标的识别。采用kalman预测滤波对脱靶量滞后时间进行补偿,选用目标空间位置进行多步预测,引导伺服机构跟踪。外场实验表明,该方法能有效她匹配识别出目标,并保持连续稳定的跟踪。     

HIV-2 Neutralization Sensitivity in Relation to Co-Receptor Entry Pathways and Env Motifs

HIV-2, compared to HIV-1, elicits potent and broadly neutralizing antibodies, and uses a broad range of co-receptors. However, both sensitivity to neutralization and breadth of co-receptor use varies between HIV-2 isolates, and the molecular background is still not fully understood. Thus, in the current study, we have deciphered relationships between HIV-2 neutralization sensitivity, co-receptor use and viral envelope glycoprotein (Env) molecular motifs. A panel of primary HIV-2 isolates, with predefined use of co-receptors, was assessed for neutralization sensitivity using a set of HIV-2 Env-directed monoclonal antibodies and co-receptor indicator cell lines. Neutralization sensitivity of the isolates was analysed in relation target cell co-receptor expression, in addition to amino acid motifs and predicted structures of Env regions. Results showed that HIV-2 isolates were more resistant to neutralizing antibodies when entering target cells via the alternative co-receptor GPR15, as compared to CCR5. A similar pattern was noted for isolates using the alternative co-receptor CXCR6. Sensitivity to neutralizing antibodies appeared also to be linked to specific Env motifs in V1/V2 and C3 regions. Our findings suggest that HIV-2 sensitivity to neutralization depends both on which co-receptor is used for cell entry and on specific Env motifs. This study highlights the multifactorial mechanisms behind HIV-2 neutralization sensitivity.     

高导电性ZAO陶瓷靶材及薄膜的制备

用热等静压法烧结制备了高导电性ZAO(铝掺杂氧化锌)陶瓷靶材,并用直流磁控溅射法制备出ZAO透明导电薄膜。靶材的致密度达98.7%,电阻率为2.2?03·cm;制得薄膜的最低电阻率为9.3?04·cm,可见光平均透射率大于85%。浅析了靶材的组织结构及靶材的电学、力学性能和薄膜制备的主要实验参数对其光、电性能的影响。     

方位-多普勒频率合成TMA的可测性分析

应用了非线性系统可测性理论对方位频率TMA的可测性进行了分析.结果表明,三维情况下在方位角和高低角的变化率满足不全为零条件时,载机不必做机动就可以实现可测.     

低合金钢靶板受尖头弹冲击的变形与断裂

研究了低合金钢板在尖头弹冲击条件下的侵彻过程,讨论了应力条件和材料冶金质量对损伤和断裂的影响。结果表明,在侵彻过程前期靶材主要表现为三向压应力下的塑性流动;中期,靶材挤压流动,靶材内部的流速差造成剪切变形,并导致沿轧制偏析带的开裂;后期,靶材在剧烈剪切作用下剪切变形、局部失稳、形成剪切带和沿剪切带的开裂或撕裂。断裂按微孔聚集型机制进行。通常所谓花瓣型开裂实质上就是上述在切应力作用下的剪切撕裂。     

基于IMM-CKF的弹道再入目标跟踪研究

该文研究了弹道系数未知的弹道再入目标的跟踪问题.针对现有再入目标跟踪方法对弹道系数初值设定以及噪声协方差的设置比较敏感的问题,采用了数值精度高、稳定性好且计算量较小的求容积卡尔曼滤波(CKF)算法作为跟踪滤波器,并分别设计了由不同弹道系数模型构成的交互式多模型(IMM)算法以及由不同噪声协方差模型组成的IMM算法.仿真结果表明,该文设计的两种IMM算法都能显著提高跟踪精度以及对弹道系数估计的收敛速度.     

刚性弹丸对多层复合靶的侵彻分析

本文研究了刚性球形弹丸对间隔靶或多层靶的侵彻问题,这些靶由不同的材料构成.给出了弹丸与靶体相互作用的模型,由它计算出了相互作用的表面在每个时间增量时的变化状态,从而可获得靶的阻力,将这些代入到弹丸的运动方程中,即可得到一个完整的弹丸在击穿靶或未击穿靶时的弹道轨迹.其中靶的阻力从广义的Poncelet方程中求得.这样,经典的静力学和动力系数与靶的材料特性就通过弹腔扩张模型联系起来,所得到的实验结果与计算的穿透结果吻合的较好.     

Quantitative measurement of oxygen content in Si targets

Enriched self-supporting targets of ^28,29,30Si are often used in the field of high resolution nuclear physics experiments. The Si targets were prepared by the thermal evaporation deposition or reduction-deposition methods. The targets, however, are always contaminated by oxygen as Si-material is easily oxidised. These contaminated targets lead to poor data with low S/N ratio. Thus it is very useful if one can estimate the amount of oxygen in the target quantitatively before experiments.In this work we have developed a method to estimate the amount of oxygen in the Si target. We used α-ray and β-ray thickness gauges, which have different sensitivity to oxygen atoms. Namely, the α-ray gauge is more sensitive to light elements such as oxygen compared to the β-ray gauge. Thus one can deduce the amount of oxygen by comparing the oxygen thickness measured by α-ray gauge with that of the β-ray gauge.Accuracy of a few percents can be obtained for the oxygen content in Si targets with a thickness of 1 mg/cm². The present method can also be applied to heavy elements such as Ta₂O₅.     

基于深度学习的采摘机器人目标识别定位算法

为了提升采摘机器人对果实的识别准确率以及定位定精度,提出一种基于深度学习Faster-RCNN框架的采摘机器人目标识别和定位算法.首先采用卷积神经网络VGG16模型提取输入图像的特性信息,并利用区域提议网络RPN生成含有目标的候选框,通过引入自适应候选框数的方法有效提升了算法性能,然后利用多任务损失函数对目标进行分类识...     

城镇居民人均全年可支配收入与主要经济指标分析

为了验证城镇居民人均全年可支配收入与地区主要经济指标是否存在正想关系,以宏统计数据库为数据源,采用2004年的统计数据,利用SAS统计软件运用回归分析等统计方法对全国31个城市的10项经济指标进行分析。研究结果表明:城镇居民人均全年可支配收入与工业、服务业、地方财政、卫生事业有线性关系,与人口、农业、固定资产投资、教育文化关系不大。