摘要
要改善高光谱图像目标检测的算法对目标先验信息的敏感度过高的问题,发现了一种偏振高光谱图像约束的能力变得最小的检测方法。从声光可调谐滤光器光谱成像仪的工作原理和系统的结构方面入手,在约束能量最小化目标检测算法中运用一种处理过的偏振高光谱数据替代掉传统的高光谱图像数据作为检测的样本,对检测结果实行连通的区域进行判断和轮廓进行提取和处理。采用AOTF偏振成像系统进行采集的偏振光谱数据来完成目标检测试验,试验结果证明,替换样本数据后的目标检测算法的精准度得到了很大程度的提高。
To improve the high sensitivity of hyperspectral image target detection algorithms to the prior information of the target,a detection method that minimizes the constraining ability of polarized hyperspectral images has been found.Starting from the working principle of the acousto-optic tunable filter spectral imager and the structure of the system,a processed polarization hyperspectral data is used to replace the traditional hyperspectral image data in the constrained energy minimization target detection algorithm.For the sample,the connected area of the detection result is judged and the contour is extracted and processed.The polarization spectrum data collected by the AOTF polarization imaging system is used to complete the target detection experiment.The experimental results prove that the accuracy of the target detection algorithm after replacing the sample data has been greatly improved.
出处
《今日自动化》
2021年第9期136-137,共2页
Automation Today
关键词
光谱成像
偏振光谱
声光可调谐滤光器
目标检测
CEM算法
spectral imaging
polarization spectrum
acousto-optic tunable filter
target detection
CEM algorithm