Investigation of the Effects of Data Collection on Visual Stylometry

Abstract

Visual stylometry is the task of analysing visual art by mathematical and statistical methods. One branch of visual stylometry is to classify if a painting/drawing is made by the claimed artist; that this is achievable has been demonstrated for several artists and by different approaches. The present authors developed a contourlet-based classification method in a previous paper and here it is investigated how robust the conclusions of this method are to variations in the data collection.