A Comceptual Structure for Computer Vision.pdfVIP

2011 Canadian Conference on Computer and Robot Vision A Conceptual Structure for Computer Vision Gregor Miller, Sidney Fels and Steve Oldridge Human Communication Technologies Laboratory ECE, University of British Columbia Vancouver, Canada gregor, ssfels, steveo@ece.ubc.ca Abstract—The research presented in this paper represents base, not just specialists in graphics. The abstraction gives several novel conceptual contributions to the computer vision programmers access to conceptual processes e.g. drawing a literature. In this position paper, our goal is to de?ne the scope textured triangle but hides the complexity of the operation of computer vision analysis and discuss a new categorisation of the computer vision problem. We ?rst provide a novel from the programmer. However, graphics is a structured decomposition of computer vision into base components which problem and is more easily represented in concept. The we term the axioms of vision. These are used to de?ne image understanding problem is unstructured and requires a researcher-level and developer-level access to vision algorithms, higher level of sophistication to extract meaningful models. in a way which does not require expert knowledge of computer Our discussion in this paper attempts to bring control of vision. We discuss a new line of thought for computer vision by basing analyses on descriptions of the problem instead of in sophisticated concepts such as image registration to a level terms of algorithms. From this an abstraction can be developed similar to that of drawing a triangle. to provide a layer above algorithmic details. This is extended The central theme of our contributions are accessibility: to the idea of a formal description language which may be we would like to see computer vision becoming more automatically interpreted thus allowing those not familiar