关键词: 水墨画, 特征提取, BP神经网络, 直方图

Abstract: In recent years, content-based retrieval and classification of nature images is widely studied, most of which take advantage of color component. Therefore, such methods are not suitable for IWPs analysis. In this paper, after image pre-processing, grayscale histogram is applied to describing the intensity distribution and compositional style. Based on Back-Propagation（BP） neural networks, Chinese arts are classified by observation and analysis. The experimental results prove that it is feasible to extract global feature based on histogram, and it can achieve high accuracy of automated classification.

Key words: Ink and Wash Paintings（IWPs）, feature extraction, BP neural networks, histogram