Small-Target Detection between SAR Images Based on Statistical Modeling of Log-Ratio Operator

Area of Science:

• Remote Sensing
• Signal Processing

Background:

• Log-ratio (LR) operators are effective for change detection in Synthetic Aperture Radar (SAR) data.
• Accurate statistical modeling of LR operators is crucial for multi-temporal SAR image analysis.

Purpose of the Study:

• To derive and parameterize the probability density function (PDF) of the LR operator for multi-temporal SAR images.
• To develop maximum-likelihood (ML) estimation for LR PDF parameters.
• To apply the proposed model to small target detection using Constant False Alarm Rate (CFAR) thresholds.

Main Methods:

• Analytical derivation of the LR operator's PDF.
• Parameterization of the LR PDF using number of looks, coherence magnitude, and true intensity ratio.
• Development of ML estimation for LR PDF parameters.
• Application to CFAR detection for small targets in SAR images.

Main Results:

• A precise statistical model for LR operator PDF was derived and parameterized.
• ML estimation methods for the LR PDF parameters were developed.
• The proposed model demonstrated effectiveness in small target detection compared to generalized Gaussian distribution.

Conclusions:

• The derived LR PDF and ML estimation provide accurate statistical models for multi-temporal SAR change detection.
• The proposed method enhances small target detection capabilities in SAR imagery.
• The model offers a valuable tool for operational SAR data analysis.