The statistical distribution of natural phenomena is of great significance in studying the laws of nature. In order to study the statistical characteristics of a random pulse signal, a random process model is proposed...The statistical distribution of natural phenomena is of great significance in studying the laws of nature. In order to study the statistical characteristics of a random pulse signal, a random process model is proposed theoretically for better studying of the random law of measured results. Moreover, a simple random pulse signal generation and testing system is designed for studying the counting distributions of three typical objects including particles suspended in the air, standard particles, and background noise. Both normal and lognormal distribution fittings are used for analyzing the experimental results and testified by chi-square distribution fit test and correlation coefficient for comparison. In addition, the statistical laws of three typical objects and the relations between them are discussed in detail. The relation is also the non-integral dimension fractal relation of statistical distributions of different random laser scattering pulse signal groups.展开更多
文摘无人机自主察打对地攻击场景中,针对无人机作战时效性强,地面目标识别场景复杂,存在模型训练、推理速度慢,小目标检测漏检、误检的问题,提出一种基于注意力机制与通道重排思想的无人机对地目标检测算法。该算法引入CA(coordinate attention)注意力机制,可提高网络对关注部分的特征提取能力;且对主干网络进行通道重排(channel shuffle)轻量化处理,可有效减少多次卷积造成的特征损失;最后,为提升战时训练及推理速度,替换部分激活函数为H-Swish,优化其损失函数为CIoU(complete intersection over union)。实验证明:采用改进的新算法,提升了28.4%训练速度,目标识别的平均精度均值(mean average precision, mAP)达99.1%,可实现最小目标检测为19*25像素,经TensorRT加速后检测速率达72.99 FPS,满足实时检测需求,针对复杂地形下的坦克小目标检测性能较好。
文摘The statistical distribution of natural phenomena is of great significance in studying the laws of nature. In order to study the statistical characteristics of a random pulse signal, a random process model is proposed theoretically for better studying of the random law of measured results. Moreover, a simple random pulse signal generation and testing system is designed for studying the counting distributions of three typical objects including particles suspended in the air, standard particles, and background noise. Both normal and lognormal distribution fittings are used for analyzing the experimental results and testified by chi-square distribution fit test and correlation coefficient for comparison. In addition, the statistical laws of three typical objects and the relations between them are discussed in detail. The relation is also the non-integral dimension fractal relation of statistical distributions of different random laser scattering pulse signal groups.
文摘激光驾束制导是一种重要制导方式,制导仪激光信息场的质量直接影响导弹制导精度。为实现快速、直观地评价激光信息场质量,提出一种基于信息场光斑图像的快速测试与处理方法。在内外场快速搭建测试系统,并对系统进行标定;基于漫反射低照度瞬态成像原理,利用低照度相机拍摄制导仪激光束,得到直观的条纹状光斑图像;通过快速图像处理技术,计算光斑尺寸、光斑中心、条纹亮度等参数来评价制导仪光轴稳定性、信息场能量均匀性等技术指标。试验结果表明:该测试方法可得到激光信息场的条纹状图像,实现对制导仪性能的快速检测,测试精度1.8 mm,处理时间0.35 s.
