Classification Of Ground Moving Object Using Coefficient Of Integrated Bispectrum For Doppler Radar

This paper considers the classification of radar target using Backscatter Doppler signature of moving object. Classification performance evaluated by the integrated Bispectrum based technique of feature extraction and compared it with Cepstrum based feature extraction technique. Classifier performance is tested by GMM (Gaussian Mixture Model) and ML (Maximal Likelihood) decision making method. Classifier is trained and tested by distinct target echoes such as single human, double human, and triple human set. Proposed classification results shows the superiority of integrated Bispectrum method over Cepstrum method and classification rates are up to 82% to 87% at different feature sets.