Edge detection techniques are some of the most useful techniques to determine the lineaments of buried structures from magnetic data. In this context, many techniques have been designed by using the magnetic gradient tensor (MGT). In this work, eigenvalue analysis of the full MGT and a new normalized approach [normalized edge detector (NK)], which balances strong and weak amplitude anomalies simultaneously, was utilized to outline lineaments of the buried archaeological structures. Satisfactory results were obtained from noisy and noise free synthetic data which were generated from prism models. The techniques were implemented on the archaeological magnetic data acquired from a buried ancient cistern. In addition, three‐dimensional (3D) electrical resistivity tomography (ERT) inversion sections and two‐dimensional (2D) magnetic inversion section and tilt angle methods were used to compare them against the results of edge detection analysis. The results of the normalized eigenvalue analysis agreed closely with all techniques in detecting the lineaments of the buried archaeological structure. NK detected the lineaments of the buried structure more clearly than the tilt angle and had better resolution. All of the results have shown that eigenvalue analysis of the full MGT and especially the newly introduced approach NK can provide important information for archaeological studies.

中文翻译：

---

利用地球物理图像处理方法对土耳其阿拉巴达州古代蓄水池的埋藏考古结构进行边缘检测

边缘检测技术是根据磁性数据确定掩埋结构的线条最有用的技术。在这种情况下，已经通过使用磁梯度张量（MGT）设计了许多技术。在这项工作中，利用了整个MGT的特征值分析和一种新的归一化方法[归一化边缘检测器（NK）]，该方法同时平衡了强振幅和弱振幅异常，勾勒出了埋藏考古结构的轮廓。从棱镜模型生成的嘈杂和无噪声的合成数据中获得了令人满意的结果。该技术是在从一个埋藏的古代水箱中获取的考古磁性数据上实施的。此外，使用三维（3D）电阻层析成像（ERT）反演部分和二维（2D）磁反演截面和倾斜角方法将它们与边缘检测分析的结果进行比较。归一化特征值分析的结果与所有用于检测埋藏考古结构轮廓的技术都非常吻合。NK比倾斜角度更清楚地检测到埋藏结构的线条，并具有更好的分辨率。所有结果表明，完整MGT的特征值分析，尤其是新近引入的NK方法，可以为考古研究提供重要信息。归一化特征值分析的结果与检测埋藏考古结构的构造的所有技术都非常吻合。NK比倾斜角度更清楚地检测到埋藏结构的线条，并具有更好的分辨率。所有结果表明，完整MGT的特征值分析，尤其是新近引入的NK方法，可以为考古研究提供重要信息。归一化特征值分析的结果与所有用于检测埋藏考古结构轮廓的技术都非常吻合。NK比倾斜角度更清楚地检测到埋藏结构的线条，并具有更好的分辨率。所有结果表明，完整MGT的特征值分析，尤其是新近引入的NK方法，可以为考古研究提供重要信息。