A Methodological Framework for Free and Open-Source UAV-Based Archaeological Research

Abstract

Full-coverage pedestrian survey to record cultural features on unexplored archaeological landscapes is costly in terms of time, money, and personnel. Over the past two decades, researchers have implemented remote sensing and landscape data collection techniques using unmanned aerial vehicles (UAVs) to combat some of these burdens, but the initial cost of equipment, software, and processing power has hindered the ubiquitous implementation of UAV technology as an accessible companion tool to traditional archaeological survey. This article presents a free and open-source, technology-independent analytical framework for the collection and processing of UAV images to produce high-resolution digital terrain models limited only by the equipment available to the researcher. Results from the free and open-source protocol are directly compared to those produced using proprietary software to illustrate the capabilities of freely available data processing tools for UAV-collected images. By replicating the methods outlined here, researchers should be able to identify and target areas of interest to increase fieldwork efficiency, decrease costs of implementing this technology, and produce high-resolution digital terrain models to conduct spatial analyses that pursue a deeper understanding of cultural landscapes.

Research on segmentation algorithm of UAV remote sensing image based on deep learning

“…Later, Li et al applied the improved FCM clustering algorithm to the segmentation of UAV remote sensing images [25]. The principle of improvement is that the number of clusters and cluster centers are no longer determined in a random way, but are determined by a histogram.…”