A Framework Towards Quantified Artistic Influences Analysis

In the past, various problems in art imaging such as painter identification, painting image classification and retrieval were successfully solved by computer vision. However, very few works focused on analyzing the relationships among paintings of different artists. In this paper, we first define a set of image features in terms of abstract artistic concepts on colour and composition. Using extracted features, we are able to classify and visualize paintings from three schools of art: renaissance, impressionism and postimpressionism in the feature space. Furthermore, we use paintings from Picasso as an example to demonstrate how artistic influential relationship is derived and visualized in the feature space. Finally, experiments show the results from our quantified artistic influences analysis are consistent with relevant information in various resources of art history.