Search

Independent Component Analysis (ICA) is a statistical technique for decomposing information from datasets into maximally independent components. ICA allows the researcher to recover two or more independent signals that appear mixed within the same dataset. This paper shows ICA to be an extremely effective method for separating different colours found in rock paintings into discrete images or components. The comparison between the results of ICA and PCA (Principal Component Analysis) shows that ICA accurately separates panels with more than one type of colour, while PCA achieves a lower degree of separation. This study also shows that in scenes with monochrome depictions, ICA tends to be slightly more effective in separating the pigments from the rock. The ICA method has been applied successfully to several rock art panels from Northern Chile, where the use of diverse types of mineral pigments is common. Two analyses conducted at the Pampa El Muerto 11 site in the Northern Chilean highlands reveal how ICA can contribute to a more compelling interpretation of more intricate panels. The comparison between the results of ICA and PCA (Principal Components Analysis) shows that ICA correctly separates panels with more than one type of pigment, while PCA achieves a lower degree of separation. This study also shows that in scenes with monochrome depictions, ICA tends to be slightly more effective in separating the pigments from the rock. ICA algorithm has been successfully in several rock panels from Northern Chile, where the use of diverse types of mineral pigments is usual. Two panels from the Pampa El Muerto site have been analysed with the technique mentioned above, informing that its application can collaborate on a more compelling interpretation of intricate panels. • A new algorithm (FastICA) for enhancing of rock paintings in digital images is analysed. • FastICA can be a reliable solution for the analysis of panels with more than one shade of pigment and superimposition of depictions. • FastICA algorithm is more efficient in separating different colours than Principal Component Analysis (PCA). • We propose a protocol for the assessment of the outcomes of different algorithms. [ABSTRACT FROM AUTHOR]