A Novel Method for Sea-Land Clutter Separation Using Regularized Randomized and Kernel Ridge Neural Networks.

Area of Science:

• Radar Signal Processing
• Machine Learning Applications
• Environmental Sensing

Background:

• Distinguishing sea clutter from land clutter is vital for shore-based radar applications.
• Traditional methods rely on clutter models and coastal maps, which have limitations.

Purpose of the Study:

• To develop and evaluate novel machine learning approaches for accurate sea-land clutter separation.
• To enhance classification performance using advanced neural network architectures.

Main Methods:

• Implementation of two neural network models: Regularized Randomized Neural Network (RRNN) and Kernel Ridge Regression Neural Network (KRR).
• Utilizing statistical features like energy variation, amplitude change frequency, and autocorrelation for classification.
• Evaluation on a mixed dataset of synthetic land clutter and real sea clutter data.

Main Results:

• The RRNN and KRR methods achieved high classification accuracies of 98.50% and 98.75%, respectively.
• Both proposed methods significantly outperformed conventional Support Vector Machine (SVM) and Extreme Learning Machine (ELM) solutions.
• Demonstrated superior performance in distinguishing sea from land clutter.

Conclusions:

• The proposed RRNN and KRR methods offer a highly accurate and effective solution for sea-land clutter separation.
• Machine learning, particularly neural networks, provides a powerful alternative to traditional methods in radar clutter classification.
• The feature engineering approach significantly contributes to the improved classification performance.