为了实现对高光谱图像中的目标自动检测,提出了一种基于空间上下文单类分类器的目标检测算法。对所采用的空间与光谱结合的特征、SVDD分类器原理、算法流程等进行研究。首先分析了支持向量数据描述(SVDD,support vector data description)的单类分类原理。接着,结合高光谱图像特点,介绍了如何利用空间上下文信息和光谱特征作为SVDD分类器输入特征。然后,在分析比较空间光谱结合单类分类器性能的基础上,说明了采用该算法的原理。最后,给出了该算法的具体实现方法。实验结果表明:该方法优于常规的
At liquid-nitrogen temperature, at 10-kHz pulse repetition rate, Q-switched 36-ns pulses with average output power of 4 W at 2.05 um and 4.5-W continuous wave output power with a total optical-optical conversion efficiency of 30%, were achievedfrom a 6% T
Structured environments are employed in a plethora of applications to tailor dynamics of light–matter interaction processes by modifying the structure of electromagnetic fields. The promising example of such a system is antiresonant photonic crystal fiber