Abstract
Conducting comprehensive excavations in the extensive areas surrounding ancient tumulus clusters can reveal the locations of undiscovered tumuli. However, such endeavors demand significant manpower, resources, and time. By preemptively estimating the locations of these tumuli, we can avoid unnecessary excavations and improve efficiency. This study aims to examine the correlation between the distribution of ancient tumuli and topographical features. Leveraging this correlation, our goal is to forecast the likelihood of encountering ancient tumuli in various unexplored areas around the Iwase Senzuka tumulus cluster. This paper proposes a novel method to analyze the topographic point cloud data. The method involves superimposing archaeological markings onto the derived topographic map for annotation, cropping out patches and combining them to predict the likelihood of tumulus presence. The generated 2D distribution heatmaps are integrated and mapped to a color-coded point cloud for simultaneous observation of the important feature distribution identified by the deep learning network and examination of the 3D topography.
Tumulus Distribution Prediction in the Iwase Senzuka Tumulus Cluster with Topographic Maps
Fullsize Image
