Line and boundary detection in speckle images

Area of Science:

• Image processing and analysis
• Speckle noise modeling
• Signal detection theory

Background:

• Speckle noise is prevalent in coherent imaging systems like SAR and ultrasound.
• Accurate line detection is vital for analyzing these images, e.g., identifying tissue boundaries.
• Existing methods may lack optimality or computational efficiency.

Purpose of the Study:

• To derive an optimal line detector for fully developed speckle noise.
• To evaluate computationally efficient suboptimal detectors.
• To explore applications in medical ultrasound imaging.

Main Methods:

• Derivation of an optimal line detector based on physical principles of speckle.
• Comparison of the optimal detector with several suboptimal, computationally efficient rules.
• Analysis of detector performance under uncorrelated and correlated (colored) speckle conditions.

Main Results:

• The optimal detector for fully developed speckle was derived.
• A simple suboptimal detector demonstrated near-optimal performance for uncorrelated noise.
• Related detectors showed potential for approaching optimal performance in colored speckle.

Conclusions:

• The derived optimal detector provides a benchmark for line detection in speckle imagery.
• Computationally efficient suboptimal methods offer practical alternatives with minimal performance loss.
• The technique is applicable to medical ultrasound for tissue boundary detection.