P. Robertson and R. Laddaga (USA)
: Information Fusion, Computer Vision, In telligent Agents, Model Based Reasoning.
Information fusion is a difficult problem in general, and an especially difficult problem for vision systems. Ev idence from multiple sources often cannot be combined in any straightforward manner, and significant reasoning and transformation operations may need to be performed on one piece of evidence before it can be usefully com bined with another. Conventional approaches to informa tion fusion are generally limited to cases where it is possi ble to combine signals at the signal processing level, before higher level interpretation takes place, or simply to process different signals independently, and then choose the "best" interpretation. Our approach combines evidence at mul tiple levels in the interpretation problem. We discuss the GRAVA architecture, and how it enables a flexible type of information fusion. The GRAVA architecture has been ap plied to several computer vision applications, and we use its application to face recognition to highlight the support for information fusion.
Important Links:
Go Back
