An ISAR object recognition system has been described. The feature extraction of ISAR object images is achieved by two-dimensional FFT processing,and a square window which is located on the center of the spectrum is used for feature selection, and the classifier of the system is a nearest neighbor classifier. Through experiments on ISAR object recognition, the effect of the feature window length on the system recognition rate has been investigated. The experimental results show that the feature selection window with the low-pass form is the optimum feature selection approach, and an optimum feature window length is existing for this feature selection method, which can be determined by training sample set itself.
展开▼
