MEAN SHIFT: A NON-PARAMETRIC ALGORITHM FOR THE SEGMENTATION OF ANOMALIES IN GEOPHYSICAL IMAGES OBTAINED FROM MAGNETIC PROSPECTION DATA

作     者：Salguero, F. Prat, F. Moreno, F. Romero, S. 

作者机构：Univ Huelva Dept Min Engn Mech & Energet Huelva 21071 Spain Univ Huelva Dept Math Huelva 21071 Spain 

出 版 物：《ARCHAEOMETRY》 (考古术)

年 卷 期：2011年第53卷第3期

页      面：642-659页

核心收录：

学科分类：0601[历史学-考古学] 06[历史学] 07[理学] 0708[理学-地球物理学] 0703[理学-化学] 0712[理学-科学技术史(分学科，可授理学、工学、农学、医学学位)] 

主　　题：MEAN SHIFT SEGMENTATION MAGNETIC PROSPECTION GILENA (SW SPAIN) DIGITAL IMAGE PROCESSING NON-PARAMETRIC ALGORITHM k-MEANS 

摘      要：This paper studies the applicability of the Mean Shift algorithm as support in interpreting geophysical images produced, on this occasion, from magnetic prospection data. The data obtained from a magnetic survey carried out in Gilena (Seville province, Spain) by the La Rabida Archaeophysics Group will be used for the research. Its applicability is illustrated by comparing, on the one hand, some reduction-to-pole algorithms and on the other, the (well-known) k-means algorithm. Finally, the paper shows the results obtained by applying the Mean Shift algorithm as an alternative method to unsupervised clustering of anomalies that appear in images obtained from geophysical data, in which the a priori knowledge of the number of classes is difficult or impossible.