Miniaturized five fundamental issues about visual knowledge
- PDF / 405,911 Bytes
- 4 Pages / 595.32 x 841.92 pts (A4) Page_size
- 95 Downloads / 202 Views
1
Frontiers of Information Technology & Electronic Engineering www.jzus.zju.edu.cn; engineering.cae.cn; www.springerlink.com ISSN 2095-9184 (print); ISSN 2095-9230 (online) E-mail: [email protected]
Perspective:
Miniaturized five fundamental issues about visual knowledge
Yun-he PAN Institute of Artificial Intelligence, College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China E-mail: [email protected]
https://doi.org/10.1631/FITEE.2040000
1 Fundamental issue 1: expression of visual knowledge Cognitive psychology has for a long time considered that an important part of human memory is visual knowledge, which is used for conducting concrete thinking. Therefore, visual-based artificial intelligence (AI) is a subject that AI cannot bypass and is of great significance. As a continuation of the previous article “On visual knowledge” (Pan, 2019), in this paper we discuss five fundamental issues about visual knowledge. Cognitive psychology experiments have been carried out to distinguish the characteristics of visual knowledge from those of language knowledge: (1) Visual knowledge can express the dimensions, color, texture, spatial shape, and spatial relationships of an object; (2) Visual knowledge can convey the movement, speed, and temporal relationships of an object; (3) Visual knowledge can facilitate space-time transformation, manipulation, and reasoning of objects (for example, the transformation of shape, action, velocity, and scene, and various spatio-temporal analogies, associations, and predictions based on spatio-temporal reasoning results). The expression of structural features realized by computer graphics (CG) is well suited to express the © Zhejiang University and Springer-Verlag GmbH Germany, part of Springer Nature 2020
manipulation and transformation of characteristics (1) and (3), but it is challenging to apply CG to simulated imaginative transformation because CG expresses in a geometric form rather than a visual concept. A visual concept should be composed of a prototype and a domain. For instance, apples have everchanging forms, but they must have one or several fundamental shapes and colors comprising what is known as a prototype. Within the prototype, a variety of apples constitute a range of transformation which has a boundary. Forms within the boundary belong to the category “apple,” beyond which they will become other fruits. This range of transformation is the scope of the concept of “apple.” Visual concepts have a hierarchical structure, which is a spatial organization containing sub-concepts. Moreover, visual concepts have an action structure, which should contain the typical motion and action categories of each subconcept in the structure. The visual proposition is the expression of the spatial and temporal relationships of a visual concept. The spatial relationship can be regarded as the scene structure, which describes locational relationships (e.g., up and down, left and right, front and rear), distance relationships, internal and external relationships, a
Data Loading...
