An assessment of methods for the digital enhancement of rock paintings: the rock art from the precordillera of Arica (Chile) as a case study. (March 2015)

• Record Type: Journal Article
• Authors: Cerrillo-Cuenca, Enrique; Sepúlveda, Marcela
• Is Part Of: Journal of archaeological science. Volume 55(2015:Mar.)
• Page Start: 197
• Page End: 208
• Publication Date: 2015-03
• Subjects: Rock art paintings -- Digital methods -- Statistics -- Principal components analysis -- Decorrelation of images; Archaeology -- Periodicals; Archéologie -- Périodiques; 930.1
• DOI: 10.1016/j.jas.2015.01.006
• Languages: English
• ISSNs: 0305-4403

Abstract

Abstract: The digital tracing of rock art is becoming a standard for archaeologists working in this field of research. The lack of specific software for this task has resulted in archaeologists either using solutions that are not statistically robust enough or working with overly generic fields of image analysis. This paper will assess the application of three techniques for digital tracing: Principal Components Analysis, K-means, and Decorrelation Stretch. In addition to these techniques of image analysis, this paper will also explore three selective techniques that classify or enhance pigmentation. These analyses have been implemented in a software package called PyDRA (developed by one of the authors, ECC). This software makes use of several scientific libraries for the digital analysis of an image. As a case study, we chose three rock art sites located between 3100 and 3500 m above sea level in the precordillera of Arica, the northern region of Chile. All of the paintings are located inside rock shelters that are easily accessible; however, we lack a systematic recording for analysing these sites. Pampa El Muerto 14 and Mullipungo 1 were recorded through direct tracings between 1980 and 1990. The Lupica 1 site was identified only in 2013 and has not been recorded until now. Due to the advancement of technology in the years since the 1980s, we have been able to compare the proficiency of different digital and statistical techniques. Our study uses the most adequate parameters and enables us to trace the paintings digitally without actually handling the surface of the rock. Highlights: New algorithms (SISC, SIVC, SISVC) to enhance rock art imagery are introduced. Algorithms for the enhancement of images (PCA, DS, SISC and variants) are discussed. Custom software (PyDRA) based on open source scientific libraries has been developed. Different Chilean rock art sites, each with its own conditions, have been analysed. Neat digital tracings for each site have been obtained through K-means classification.