Remote hyperspectral imagery as a support to archaeological prospection
Remote hyperspectral imagery as a support to archaeological prospection
复制标题
DOI：
10.1016/j.culher.2007.03.003
复制
发表时间：
2007-07-01
影响因子：
3.1
通讯作者：
Poscolieri, Maurizio
中科院分区：
文献类型：
--
作者：
Cavalli, Rosa Maria;Colosi, Francesca;Poscolieri, Maurizio
利用高光谱数据来测试其作为考古勘探工具的有效性，并设想它们在检测与埋藏考古结构相关的光谱异常方面的潜力。为此，对机载多光谱红外和可见光成像光谱仪图像进行了分析。对整个数据集的每个单频带和不同的处理技术产品进行解释，以识别任何音调异常。由于每个分析的图像在大小和强度方面都表现出不同的标记，因此定义了两个指数来评估每个图像异常检测的潜力。这些参数是：检测指数，用于计算每个图像中与标记相关的像素数量，以及可分离性指数，用于测量标记相对于背景的色调差异。这些指数在塞利农特考古公园内的两个区域进行了测试，考古学家推测那里存在尚未挖掘的遗骸。对于测试地点，任何提取的异常现象均由专家进行评估，以确定其考古相关性。对来自光谱仪的每个单波段和不同图像处理副产品的指数值进行比较，可以确定哪个光谱范围和哪种处理方法对于快速突出异常最有价值。分析指出，在植被覆盖占主导地位的情况下，可见近红外光谱区域对与埋藏结构相关的光谱特性的变化更敏感，而在土壤覆盖变得相关的情况下，短波红外和热红外区域更加敏感。就应用的处理方法而言，光谱角度映射器分类器和最小距离算法强调最高的考古信息含量。这项工作的结果表明，通过分析所应用的图像处理技术的输出而得出的考古信息内容比通过解释每个单一波段和可用的历史航空照片所获得的信息更重要。最后，应用于塞利农特考古公园整个远程高光谱数据集的数据处理流程图对于检测与埋藏考古结构的存在相关的异常现象似乎令人鼓舞。 (c) 2007 年 Elsevier Masson SAS。版权所有。
Hyperspectral data were exploited to test their effectiveness as a tool for archaeological prospection, envisaging their potential for detecting spectral anomalies related to buried archaeological structures. For this purpose the airborne Multispectral Infrared and Visible Imaging Spectrometer images were analysed. Each single band of the entire data set and different processing technique products were interpreted to identify any tonal anomalies. Since every analysed image exhibited marks different in terms of size and intensity, two indexes were defined for assessing the potential of anomalies detection of each image. Such parameters were: the Detection Index, used for counting the number of pixels related in each image to marks, and the Separability Index, applied for measuring the tonal difference of the marks with respect to the background. These indexes were tested on two areas within the Selinunte Archaeological Park where the presence of remains, not yet excavated, was supposed by archaeologists. For the test sites any extracted anomalies were evaluated by an expert in order to determine their archaeological relevance. The comparison among the index values, derived from each single band of the spectrometer and from different image processing by-products, allowed to determine which spectral range and which processing method are the most valuable to quickly highlight the anomalies. The analysis pointed out that, where vegetation cover is dominant, the Visible near infrared is the spectral region more sensitive to variations of spectral properties related to buried structures, while, where soil cover becomes relevant, the Short-wave infrared and the Thermal-infrared regions resulted more sensitive. As far as the applied processing methods are concerned, the Spectral Angle Mapper classifier and, secondly, the Minimum Distance algorithm stressed the highest archaeological information content. The results of this work showed that the archaeological information content derived by analysing the outputs of the applied image processing techniques is more significant than the information obtained by interpreting each single band and the available historical aerial photos. As a final remark, the data processing flow chart, applied to the entire remote hyperspectral data set over Selinunte Archaeological Park, appeared encouraging for detection of anomalies related to the presence of the buried archaeological structures. (c) 2007 Elsevier Masson SAS. All rights reserved.