Computational visualistics


The term Computational visualistics is used for addressing the whole range of investigating pictures scientifically "in" the computer.

Overview

take a rather prominent place in contemporary life in the western societies. Together with language, they have been connected to human culture from the very beginning. For about one century – after several millennia of written word's dominance – their part is increasing again remarkably. Steps toward a general science of images, which we may call 'general visualistics' in analogy to general linguistics, have only been taken recently. So far, a unique scientific basis for circumscribing and describing the heterogeneous phenomenon "image" in an interpersonally verifiable manner hasstill been missing while distinct aspects falling in the domain of visualistics have predominantly been dealt with in several other disciplines, among them in particular philosophy, psychology, and art history. Last, important contributions to certain aspects of a new science of images have come from computer science.
In computer science, too, considering pictures evolved originally along several more or less independent questions, which lead to proper sub-disciplines: computer graphics is certainly the most "visible" among them. Only just recently, the effort has been increased to finally form a unique and partially autonomous branch of computer science dedicated to images in general. In analogy to computational linguistics, the artificial expression computational visualistics is used for addressing the whole range of investigating scientifically pictures "in" the computer.

Areas covered

For a science of images within computer science, the abstract data type »image« stands in the center of interest together with the potential implementations. There
are three main groups of algorithms for that data type to be considered in computational visualistics:

Algorithms from »image« to »image«

In the field called image processing, the focus of attention is formed by the operations that take one picture and relate it to another picture. With these operations, we can define algorithms for improving the quality of images, and procedures for extracting certain parts of an image or for stamping out pictorial patterns following a particular Gestalt criterion. Compression algorithms for the efficient storing or transmitting of pictorial data also belong into this field.

Algorithms from »image« to "not-image"

Two disciplines share the operations transforming images into non-pictorial data items. The field of pattern recognition is actually not restricted to pictures. But it has performed important precursory work for computational visualistics since the early 1950s in those areas that essentially classify information in given images: the identification of simple geometric Gestalts, the classification of letters, the "seeing" of spatial objects in the images or even the association of stylistic attributes of the representation. That is, the images are to be associated with instances of a non-pictorial data type forming a description of some of their aspects. The neighboring field of computer vision is the part of AI in which computer scientists try to teach – loosely speaking – computers the ability of visual perception. Therefore, a problem rather belongs to computer vision to the degree to which its goal is "semantic", i.e., the result approximates the human seeing of objects in a picture.

Algorithms from "not-image" to »image«

The investigation of possibilities gained by the operations that result in instances of the data type »image« but take as a starting point instances of non-pictorial data types is performed in particular in computer graphics and information visualization. The former deals with images in the closer sense, i.e., those pictures showing spatial configurations of objects in a more or less naturalistic representation like, e.g., in virtual architecture. The starting point of the picture-generating algorithms in computer graphics is usually a data type that allows us to describe the geometry in three dimensions and the lighting of the scene to be depicted together with the important optical properties of the surfaces considered. Scientists in information visualization are interested in presenting pictorially any other data type, in particular those that consist of non-visual components in a "space" of states: in order to do so, a convention of visual presentation has firstly to be determined – e.g., a code of colors or certain icons. The well-known fractal images form a borderline case of information visualization since an abstract mathematical property has been visualized.

Computational Visualistics Degree Programmes

The subject of computational visualistics was introduced at the University of Magdeburg, Germany, in the fall of 1996. It was initiated by Thomas Strothotte, Prof. for computer graphics in Magdeburg and largely supported by Jörg Schirra together with a whole team of interdisciplinary researchers from the social and technical sciences as well as from medicine.
This five-year diploma programme has computer science courses as its core: students learn about digital methods and electronic tools for solving picture-related problems. The technological areas of endeavour are complemented by courses on pictures in the humanities. In
addition to learning about the traditional contexts of using pictures, students intensively practice their communicative skills. As the third component of the program, an application subject such as biology and medicine gives students an early opportunity to apply their knowledge in that they learn the skills needed for co-operating with clients and experts in other fields where digital image data are essential, e.g. microscopy and radiologic image data in biology and medicine. Bachelor and Master's programmes have been introduced in 2006.
The expression 'computational visualistics' is also used for a similar degree programme of the University at Koblenz-Landau.