A Painting Artist Recognition System Based on Image Processing and Hierarchical SVM

Over the past years, forgery paintings of famous artists have been sold as original. In order to spot the fake painting, the experts make the decision based on personal experience and with the help of examining some characteristics of the painting and painter. Applying the image processing methods to artwork can reduce the need of the expert and provide quick and reliable results to recognize the originality of the artwork. In this paper, the proposed method is able to identify the painter of artwork using image processing and data mining techniques. The method consists of two typical main stages, feature extraction, and classification. In the feature extraction, 11 statistical features are extracted from each image. These features have been selected in such a way that maximize the distinction of painters. In the second step, the painters are identified by hierarchical classification. In order to evaluate the performance of proposed method, it has applied to a collection of 348 paintings from eight Iranian artists. The method has been able to identify the artwork painter with the accuracy about 84.21%.