Aircraft recognition system using eigenvector technique

Abstract

The general task of the aircraft identification system is to select the important aspects and ignore the irrelevant data. The perimeter of the aircraft in the image plane possesses sufficient information for recognition. The contour of each aircraft was extracted using a morphological module and the perimeter was standardised in the gray level. The eigenvector technique was exploited to align the input image to a fixed orientation. The principal component analysis is used to create a new co-ordinate system for object representation where each aircraft is projected onto that plane and the projection coefficients are used for the purpose of recognition. The new features are invariant to translation, scale and rotation. The presented algorithm was tested and it was found that the recognition performance is 100% using only an x-l features where x is the number of the referenced aircraft.