Abstract: In the history of art, the analysis and demarcation of visual art style was judged by the subjective awareness of art critics. But each person has his own views, so it is too subjective. Computer, as a general-purpose digital processing machine, can count the characteristics of art style objectively. This paper uses computer as the calculation tool, uses the information theory as a theoretical basis, and proposes four image features for the art style analysis. They are palette entropy, redundancy, degree of order and complexity. This paper studies the paintings selected from seven artists in experiments. It is found that the four features can reflect the intrinsic link of the same kind paintings and the differences of the different kind paintings by the quantitative analysis and statistics of the feature values. The experimental results can echo with the analysis of critics.

Key words: visual art, painting style, information theory, degree of order, complexity

摘要： 在艺术史上，对于视觉艺术风格的分析与界定长期以来都是靠艺术评论家们的主观意识来判断，而不同的人就会有相差甚远的看法，因此存在着很大的主观性。计算机作为一个通用的数字处理机器，能够客观地统计出某些能表现艺术作品风格的特征。以计算机为计算工具，以信息论知识为理论依据，提出了图像调色板熵、冗余度、有序度、复杂度等四个能表征艺术作品风格的特征值，并选取了七位具有代表性的中西方画家的绘画作品作为研究对象，通过对所选作品的各特征值量化统计与分析，发现这些数字特征能够较好地反映出同种画派之间的内在联系和不同画派间的差别，并且能够与评论家们的分析相呼应。

关键词: 视觉艺术, 画派风格, 信息论, 有序度, 复杂度