Characterization of Rembrandt self-portraits through digital-chiaroscuro statistics

Abstract

Abstract In order to develop an analytical tool to aid connoisseurs in authentication of paintings we introduce a digital computational technique based on luminosity histograms of high-resolution digitized images of paintings. This exploratory approach emerged from the widely-held hypothesis that every individual artist (master, student, copier, or forger) will tend to have a personal hand/eye spatial “signature” in shading and contrast (e.g., chiaroscuro and sfumato) that may be revealed and quantified by means of histogram statistics. A preliminary database library is built of amplitude histogram data of the digital images of authenticated works by the Dutch painter Rembrandt van Rijn against which copies and forgeries then are compared. It illuminates the individual chiaroscuro or spatial-blending characteristics of the master-created paintings to be analyzed. We show preliminary results from our analyses of 36 self-portraits that at various times have been attributed to Rembrandt and were chosen by us as a consequence of the prodigious historical, scientific, and stylistic scrutiny that have focused on these works in recent decades. Our statistical results demonstrate probative conformity for paintings that have been established as authentic self-portraits by Rembrandt and reveal statistical differences when compared to copies and forgeries as well as paintings of the Rembrandt workshop.