Adaptive polarimetric sensing for optimum radar signature classification using a genetic search algorithm.

Area of Science:

• Radar polarimetry
• Signal processing
• Machine learning

Background:

• Radar polarimetric pattern classification is crucial for target identification.
• Optimizing polarization sensing angles can improve classification accuracy.
• Existing methods may lack adaptability and automation.

Purpose of the Study:

• To develop an automated technique for adaptive radar polarimetric pattern classification.
• To identify optimal transmit and receive polarization sensing angles.
• To evaluate the effectiveness of various pattern separation distance functions.

Main Methods:

• Utilized a genetic algorithm for optimization.
• Employed seven probabilistic pattern separation distance functions (Rayleigh quotient, Bhattacharyya, divergence, Kolmogorov, Matusta, Kullback-Leibler, Bayesian probability of error).
• Applied the method to real, fully polarimetric synthetic aperture radar target signatures, represented as functions of polarization angles.

Main Results:

• A unique set of optimal polarization angles was identified for most distance functions.
• The identified angles demonstrated potential for improved classification performance.
• The automated approach proved effective in optimizing classification parameters.

Conclusions:

• The developed automated technique successfully optimizes polarization sensing angles for adaptive radar polarimetric pattern classification.
• This optimization leads to enhanced target signature classification.
• The findings suggest a significant improvement in radar target recognition capabilities.