A Sea-Sky Line Detection Method for Unmanned Surface Vehicles Based on Gradient Saliency

Area of Science:

• Robotics and Autonomous Systems
• Computer Vision
• Marine Navigation

Background:

• Optical image interference from cloud clutter, sea glint, and weather conditions hinders accurate sea-sky line (SSL) detection for unmanned surface vehicles (USVs).
• Robust SSL detection is critical for navigation, perception, and operational safety of USVs in real marine environments.

Purpose of the Study:

• To develop a robust and accurate saliency-based method for sea-sky line (SSL) detection in challenging marine optical images.
• To enhance the performance of USVs by improving the reliability and real-time capability of SSL detection.

Main Methods:

• Gradient saliency computation to enhance SSL features and suppress interference.
• Region growing on gradient orientation to obtain line support regions.
• SSL identification using region contrast, line segment length, and orientation features.
• Cubature Kalman Filter (CKF) for optimal state estimation and improved accuracy/stability.

Main Results:

• The proposed saliency-based method effectively enhances sea-sky line features while suppressing environmental interference.
• Experimental results on a benchmark dataset show superior accuracy and real-time performance compared to state-of-the-art methods.
• The integration of the Cubature Kalman Filter significantly improved the accuracy and stability of SSL detection.

Conclusions:

• The developed saliency-based SSL detection method offers a significant advancement for USV navigation in complex marine conditions.
• The method demonstrates high accuracy, stability, and real-time performance, making it suitable for practical USV applications.
• The study validates the effectiveness of combining saliency-based image processing with advanced filtering techniques for robust environmental perception.