Automated Extraction of Archaeological Traces by a Modified Variance Analysis

Abstract

This paper considers the problem of detecting archaeological traces in digital aerial images by analyzing the pixel variance over regions around selected points. In order to decide if a point belongs to an archaeological trace or not, its surrounding regions are considered. The one-way ANalysis Of VAriance (ANOVA) is applied several times to detect the differences among these regions; in particular the expected shape of the mark to be detected is used in each region. Furthermore, an effect size parameter is defined by comparing the statistics of these regions with the statistics of the entire population in order to measure how strongly the trace is appreciable. Experiments on synthetic and real images demonstrate the effectiveness of the proposed approach with respect to some state-of-the-art methodologies.

Auto-Extraction of Linear Archaeological Traces of Tuntian Irrigation Canals in Miran Site (China) from Gaofen-1 Satellite Imagery

“…Several methods have been proposed for the semi-automatic extraction of archaeological features from aerial, satellite multispectral, and panchromatic imagery [36][37][38][39][40][41][42][43][44][45][46]. D'Orazio et al [41] proposed a semi-automatic approach for crop-mark extraction using a region-based active contour model [42].…”

“…Luo et al [45] proposed an automatic approach to perform extraction of Qanat tops using edge detection algorithm and circular Hough transformation from Google Earth imagery. D'Orazio et al [46] designed an automatic approach for crop-mark extraction using modified variance analysis to improve extraction accuracy for aerial imagery.…”

Integrating Remote Sensing, Machine Learning, and Citizen Science in Dutch Archaeological Prospection

“…Their continuous success is due to the fact that simple geometric shapes such as circles and rectangles are common in the archaeological record but rare in nature [23]. Knowledge-based algorithms (e.g., [31][32][33][34][35]) require detailed knowledge about the expected objects and their surroundings. They tend to be highly case-specific and are rarer in archaeology.…”

Global-Scale Archaeological Prospection using CORONA Satellite Imagery: Automated, Crowd-Sourced, and Expert-led Approaches

“…There is tradition of research in archaeological remote sensing dating back to the 1980s, when multispectral satellite images from the Landsat program became available, that sought to identify archaeological sites based on their spectral signatures in imagery (e.g., Limp 1989) or to use reflectance and other spatial data to construct rudimentary predictive models that would indicate where archaeological sites would be most likely discovered (e.g., Custer et al 1986). In many respects, to have a computer automatically identify sites alongside other environmental features is something of a Holy Grail of archaeological remote sensing, and recent years have seen rapid advancements in the sophistication of such efforts (e.g., Cerrillo-Cuenca 2016;Cowley 2012;D'Orazio et al 2015;Freeland et al 2016;Megarry et al 2016). Most automated detection efforts rely either on an assumption that archaeological sites will possess spectral reflectance characteristics that are sufficiently unique so as to permit them to be recognized, or alternatively, they rely on a shape-or object-based analysis, which often combines some expected range of reflectance values with an expected shape and size ( Menze and Ur's (2012) study in the Upper Khabur basin of eastern Syria provides an good example of a relatively successful automated site detection project in which the team cleverly combine a time series of Landsat and Aster images to build spectral classification for known sites, helping to highlight the anthropogenic soils that are characteristic of the region.…”