Automatic Target Recognition and Identification for Military Ground-to-Air Observation Tasks using Support Vector Machine and Information Fusion

Abstract

Automatic Target Detection, Recognition, and Identification (ADTRI) is an important task, especially for the military. This paper enhances the military technique for recognizing and identifying air objects by utilizing a Support Vector Machine (SVM) combined with information fusion. For this purpose, SVM, as the recognizer, generated knowledge of 11 characteristics that consist of the Wing, Engine, Fuselage, and Tail (WEFT) of 155 military and civilian aircraft and helicopters. Then, the identification is carried out by the information fusion from the SVM result. Using the 80:20 scheme, the combination of SVM and information fusion can achieve an average accuracy of 99.60% during training and 98.39% during the testing for the combination of primary and secondary characteristics. Another important thing is that this combination can speed up the process of recognition and identification by up to 0.52 seconds.