Image Feature Extraction and Recognition of Abstractionism and Realism Style of Indonesian Paintings

Abstract

This paper chooses and evaluates three feature vectors and their augmented feature vector for recognizing the styles of Indonesian paintings. The three feature extraction methods include the Gabor wavelet, histogram analysis, and number-of-edge analysis. The recognition purpose is to discriminate between the abstractionism and the realism styles of Indonesian paintings. The experimental results using 115 painting images show that the use of number-of-edge features has given the best result with 66.23% accuracy.