Reasoning with geographic knowledge in satellite image understanding
Improved procedures are needed to extract optimal spatial information from images provided by space-borne sensors. Neural networks can be used to provide initial classifications of imagery which can then be refined in a post-classification process by knowledge-based systems. Digital cartographic data in geographical information systems can be used in conjunction with context rule bases to facilitate image understanding to improve spatial products. Such methods require evidential reasoning procedures to handle uncertainty and lack of information. The Dempster-Shafer approach shows considerable promise for this task.
Bibliographic Reference: Paper presented: Workshop on New Tools for Spatial Analysis, Lisboa (PT), November 17-20, 1993
Availability: Available from (1) as Paper EN 38004 ORA
Record Number: 199311625 / Last updated on: 1994-11-28
Original language: en
Available languages: en