Classifying traditional Chinese painting images

More and more traditional Chinese painting art images are digitalized and exhibited on the Internet. Effective browsing and retrieving them is an imperative problem need to be addressed. This paper proposes a scheme to classify traditional Chinese paintings. The algorithm uses three low-level features to achieve such a high-level classification: Ohta histogram, color coherence vector and auto-correlation. An accuracy of 97.21% was achieved on the database of 1254 Chinese painting database.