The Classification of Style in Fine-Art Painting Thomas Lombardi Pace University [email protected] ABSTRACT The computer science approaches to the classif ication of painting concentrate primarily on painter identific ation. While this goal is certainly worthy of pursuit, there are other valid tasks related to the classification of painting including the description and analysis of the relationships between different painting styles. This paper proposes a general approach to the class ification of style that supports the following tasks: recognize painting styles, identify key relationships between styles, outline a basis for determining style proximity, and evaluate and visua lize classification results. The study reports the results of a review of f eatures currently applied to this domain and supplements the review w ith commonly used features in image retrieval. In part icular, the study considers several color features not applied to painting classification such as color autocorrelograms and d ynamic spatial chromatic histograms. The survey of color features revealed that preserving frequency and spatial information of the color content of a painting did not improve classification accura cy. A palette description algorithm is proposed for describing th e color content of paintings from an image’s color map. The palett e description algorithm performed well when compared to similar c olor features. The features with the best performance were te sted against a standard test database composed of images from the Web Museum[16]. Several supervised and unsupervised te chniques were used for classification, visualization, and ev aluation including k-Nearest Neighbor, Hierarchical Clusteri ng, Self- Organizing Maps, and Multidimensional Scaling. Sty le description metrics are proposed as an evaluation t echnique for classification results. These metrics proved to be as reliable a basis for the evaluation of test results as compara ble data quality measures. 1. INTRODUCTION Researchers are marshalling advances in digita l image processing, machine learning, and computer vision t o solve problems of the attribution and interpretation of f ine-art paintings[5,7,8,9,10,12,14,15,17,18,21,22]. The re search to date focuses on painter identification (attribution) and authentication and therefore stresses high degrees of accuracy on small target datasets. As a result of this focus, the problem o f the broad classification of style in painting receives relati vely little attention[5]. In particular, the following questio ns of style classification in painting are as yet only partiall y addressed: Is it possible to classify paintings in general way? What features are most useful for painting classification? How are th ese features different from those used in image retrieval if at all? How are style classifications best visualized and evaluated ? In answering these questions, this work endeavors to show that t he style of fine- art paintings is generally classifiable with semant ically-relevant features. Previous approaches to style classification re veal five trends in the literature. First, the solutions proposed are often style-specific addressing only particular kinds of art or even the work of particular painters[7,10,12,15,18,21]. Second, the literature emphasizes texture features while minimizing the po tential role of color features[5,14]. Third, the studies to date d o not examine techniques for evaluating classification accuracy. Fourth, current research disregards the semantic relevance of the f eatures studied[9]. Fifth, the projects currently undertak en forego a broad approach to style preferring small focused studies of particular painters or movements[18,21,22]. In contrast to previous approaches, this paper considers the components necessary to classify style in a general way with techniques that apply to a broad range of painting styles. Section 2 outlines the basis of formal approaches to painti ng style and discusses the formal elements considered in this pa per: light, line, texture, and color. In Section 3, the feature surv ey addresses feature extraction, normalization, and comparison. A palette description algorithm is defined with some addition al discussion of color features.. Section 4 reviews the classifi cation methods for several supervised and unsupervised techniques including k- Nearest Neighbor (kNN), Hierarchical Clustering, Se lf- Organizing Maps (SOM), and Multidimensional Scaling (MDS). Section 5 organizes and summarizes the results of t his paper and presents two approaches to the evaluation of classi fication results. Section 6 reiterates the conclusions of the study. 2. FORMAL APPROACHES TO STYLE The formal approach to style presupposes that art is best understood in formal terms like line, color, and sh ape rather than content or iconography. For two reasons, the forma l approach to style offers the best starting point for the comput ational classification of style in painting. First, the fo rmal elements of a painting like line and color are precisely the qual ities of images that computers can measure. Computer approaches ba sed on iconography cannot be undertaken until computer tec hniques exist to recognize objects of interest in the art domain. That is to say, until object recognition algorithms can identify a woman holding a plate adorned with two eyes, a common iconographi c representation of Saint Lucy, computer approaches t o style based on content are not feasible. Second, many styles o f painting, such as abstract expressionism, do not contain explicit identifiable content. Therefore, approaches to style based on c ontent cannot address works of art whose content is largely and e xplicitly formal. Art historians and critics use a nuanced vocab ulary to discuss the formal characteristics of paintings[1,20]. The formal terms for describing a painting focus on how an artist painte d the given subject in a particular context. Color, line, ligh t, space,
Posted on: Mon, 24 Jun 2013 18:13:28 +0000
Trending Topics
Recently Viewed Topics
© 2015