Automatic recognition of ground radar targets based on target RCS and short time spectrum variance

This paper presents a novel feature vector to be used with a robust automatic target recognition (ATR) classifier designed for a ground surveillance radar. A three element feature vector has been used where features are based on radar audio signal of 100 milliseconds duration. The short feature length allows fast real-time implementation of the classifier. Classification is done using a k-nearest neighbor (k-NN) classifier. There are two input classes to the classifier; automobile and pedestrian. Training has been done on real radar data. Classifier performance has been tested using test data from two different data sets. It has been demonstrated that in both cases, the overall classifier performance is above 80%.