Automatic Target Classification

We are working on automatic target recognition problem for ground surveillance radar. The goal of this research is to develop an automatic target classifier that can classify ground targets. The work is based on extracting the micro-Doppler features from the backscattered signal using signal processing and then classifying them using machine learning techniques.

In addition, we are developing high resolution 2D time-frequency representation of pedestrian micro-Doppler signature.

Scatter plot showing first 3 feature vectors
Scatter plot showing first 3 feature vectors

This work has resulted in a number of publications :

Machine Learning:

Joint time-frequency Representation:

Pedestrian modified 2D TDMV
  • S. Liaqat , M. B. Ihsan, S. Z. Asghar , A. Ejaz, A. Javed, “́High resolution 2D time frequency representation of radar micro-Doppler pedestrian signal”, Radar Conference (EuRAD), 2013, 10th European, Oct. 9 2013-Oct. 11 2013, pp 168 - 171.