Statistical algorithms for target detection in coherent active polarimetric images

Area of Science:

• Optical Engineering
• Remote Sensing
• Signal Processing

Background:

• Polarimetric imaging offers unique capabilities for target detection by analyzing light polarization.
• Detecting small targets in complex environments is challenging due to factors like illumination nonuniformity.
• Existing methods often rely on backscattered intensity, which can be unreliable in active polarimetric systems.

Purpose of the Study:

• To develop a robust small-target detection algorithm for polarimetric imagers.
• To overcome limitations of existing methods by avoiding reliance on backscattered intensity images.
• To create a detection algorithm that is independent of spatial illumination variations.

Main Methods:

• A novel algorithm based on nonlinear pointwise transformation of orthogonal state contrast images.
• Application of a maximum-likelihood algorithm optimized for additive Gaussian perturbations.
• Validation using both simulated and real-world polarimetric image data.

Main Results:

• The proposed algorithm demonstrates efficient small-target detection capabilities.
• The technique proves effective even without prior knowledge of scene characteristics.
• Performance is validated against optimal algorithms, showing advantages in practical scenarios.

Conclusions:

• The developed algorithm provides a simple and efficient solution for small-target detection in polarimetric imaging.
• This approach enhances target detection robustness by utilizing orthogonal state contrast images.
• The findings have implications for various applications requiring sensitive and reliable target identification.