Automatic aircraft recognition is very complex because of clutter, shadows, clouds, self-occlusion and degraded imaging conditions. This paper presents an aircraft recognition system, which assumes from the start that the image is possibly degraded, and implements a number of strategies to overcome edge fragmentation and distortion. The current vision system employs a bottom up approach, where recognition begins by locating image primitives (e.g., lines and corners), which are then combined in an incremental fashion into larger sets of line groupings using knowledge about aircraft, as viewed from a generic viewpoint. Knowledge about aircraft is represented in the form of whole/part shape description and the connectedness property, and is embedded in production rules, which primarily aim at finding instances of the aircraft parts in the image and checking the connectedness property between the parts. Once a match is found, a confidence score is assigned and as evidence in support of an aircraft interpretation is accumulated, the score is increased proportionally. Finally a selection of the resulting image interpretations with the highest scores, is subjected to competition tests, and only non-ambiguous interpretations are allowed to survive. Experimental results demonstrating the effectiveness of the current recognition system are given.