古遗址目标提取方法研究

目的研究根据获取的遥感信息从背景"影"像中有效识别和提取文物遗存的方法。方法分析研究区遗址影像波谱特征基础上,运用高光谱分析技术获得已知遗址目标波谱曲线,最终利用约束能量最小化(CEM)算法,从背景地物中分离遗址目标。结果当遗址与周围背景物性差异明显,引起明显的光谱异常时,CEM算法能够基于真实的物理模型进行波谱提取、识别和探测古遗址目标。结论 CEM算法提取古遗址信息是可行性强的一种方法,且识别效果与MNF变换后图像的维数有关,选择特征值较大的波段组合参与CEM运算,能有效地避免噪声对目标探测的干扰,使得探测率提高。

The extraction methods of the historical site targets

Aim To extract the physical property of the remains effectively from the background image according to the remote sensing information acquired.Methods Based on analyzing the spectral features of the site on image in the research area,hyperspectral analysis technology was used to get the target spectrum curve,eventually,remain targets were isolated from background features by constrained energy minimization method(CEM).Results CEM based on the physical model can be valid for spectrum extraction,historical site target identification and detection when historical site,whose spectral features are discernible from their background,can cause significant spectrum abnomal.Conclusion CEM algorithm combined with the hyperspectra analysis is an effective method to extract information of historical sites,and the effect of detection is relevant to the MNF dimensions,choosing the bands combination,whose characteristic value is larger,to participate the operation of CEM,can effectively avoid noise to the target detection of interference and improve the detection.