INTERPRETATION IN REMOTE SENSING AS ABDUCTIVE INFERENCE

 

D.Ali Algarni

King Saud University, Civil Engineering Department

P.O.Box 800, Riyadh 11421, Saudi Arabia

 

ABSTRACT

The task of image interpretation is so complicated that one or two data attributes may not be enough to obtain a reliable solution. Therefore, data of a different nature must be accommodated and utilized. However, handling all data at once in a one-step solution can be extremely difficult. Abduction, or the inference to the best explanation, is the inference that goes from data describing something to an explanatory hypothesis that best explains or accounts for the data. This concept was found to be applicable to tasks similar to that of image interpretation such as disease diagnosis, speech recognition and many other disciplines. In this paper, the image interpretation as an abductive task is introduced and analyzed. Furthermore, a model of layered abduction is presented whereby data attributes such as DEM, reflectance, brightness value, edges, texture, etc. are all incorporated into the model. These layers provide different information and consequently suggesting different treatment in evaluation. The model is a bottom-up approach in which the hypotheses of a bottom layer become data to be explained in a higher layer leading to the most plausible hypotheses that account for the collected evidence.

 

 

ACSM/ASPRS Annual Convention & Exposition. Baltimore,USA, 1994. 2:289-298.