Research on Underwater Target Detection Method Based on APO-DBSCAN Clustering.

Area of Science:

• Geophysics
• Signal Processing
• Computational Intelligence

Background:

• Traditional quality control for discrete Euler solutions in magnetic target detection suffers from excessive filtering, subjective parameter settings, and poor noise resistance.
• Existing methods struggle to reliably distinguish valid solutions from noise, impacting localization accuracy.

Purpose of the Study:

• To develop a novel, robust quality control method for discrete Euler solutions in underwater magnetic target detection.
• To enhance the precision and accuracy of magnetic target localization by optimizing Euler solution selection.

Main Methods:

• A new objective function was designed incorporating Euler solution residual penalty terms and contour line coefficients.
• The Artificial Protozoa Optimizer (APO) algorithm was employed to identify optimal DBSCAN clustering parameters.
• The APO-DBSCAN approach was validated through simulations and field tests.

Main Results:

• The optimized algorithm demonstrated a significant increase in valid solution retention rates: 52.52% for noise-free and 76.33% for noisy data.
• Invalid solution retention was substantially reduced by 28.57% (noise-free) and 94.21% (noisy data).
• Field tests showed a 28.06% reduction in the average deviation from the true center of gravity for magnetic targets.

Conclusions:

• The proposed APO-DBSCAN method offers a superior alternative to traditional statistical screening for quality control of Euler solutions.
• This approach significantly improves the reliability and accuracy of underwater magnetic target detection and localization.
• The method exhibits robust performance across varying noise levels, enhancing practical applicability.