Morphology/Fractal-Based Techniques for Automatic Target Discriminatio

This proposal describes an innovative approach to the detection and discrimination of targets in multi-sensor airborne platforms. The major advantages of our process are: 1) substantial performance gains; 2) the use of a common architecture for multi-sensor platforms; and 3) the need for less hardware for real-time implementation than current state-of-the-art techniques. Our proposed target detection and discrimination process combines two key technologies - morphological signal processing and fractal analysis. We will use geometric constraints (shape) as a first order filter for discriminating man-made objects from the natural cluttered background. This will be followed by a second step of processing (false alarm reduction and detection prioritization) which will exploit the difference in the fractal behavior of natural clutter vs. man-made objects. Both stages of the algorithm will be implemented using non-linear morphological filters, which require many fewer numerical operations than conventional signal processing algorithms. As a result we will achieve high computational efficiency and increased processing speed. For Phase I demonstration we will focus on one aspect of the Theater Missile Defense scenario. This approach offers a fast, low-cost, low-power common hardware implementation for the detection and discrimination of targets in multi-sensor data.