Speckle Noise Filtering in Side-Scan Sonar Images Based on the Tucker Tensor Decomposition

Area of Science:

• Signal Processing
• Image Analysis
• Geophysics

Background:

• Real-world signals, including sonar data, are often degraded by noise.
• Noise in signals negatively impacts the performance of signal processing systems.
• Speckle noise is a common issue in side-scan sonar imagery, hindering accurate interpretation.

Purpose of the Study:

• To develop an advanced method for speckle noise reduction in side-scan sonar images.
• To leverage tensor decomposition for enhanced noise filtering capabilities.
• To provide a robust solution for improving the quality of sonar data.

Main Methods:

• A novel tensor-based approach utilizing Tucker decomposition.
• Automatic determination of ranks for factoring tensors within the decomposition.
• Application of the method to reduce speckle noise in side-scan sonar images.

Main Results:

• The proposed tensor-based method demonstrates significant speckle noise reduction.
• Experimental verification confirms the effectiveness of the technique.
• The method outperforms conventional speckle noise filters in terms of performance.

Conclusions:

• The tensor-based Tucker decomposition offers a superior approach to speckle noise reduction in sonar images.
• This method enhances the reliability and utility of side-scan sonar data.
• The findings suggest broader applicability in signal processing where noise reduction is critical.