Complex wavelet structural similarity: a new image similarity index.

Area of Science:

• Image processing
• Computer vision
• Signal analysis

Background:

• Assessing image similarity is crucial for various applications.
• Existing image similarity indices often require preprocessing and are sensitive to distortions.
• There is a need for robust, efficient, and general-purpose image similarity metrics.

Purpose of the Study:

• Introduce the complex wavelet structural similarity (CW-SSIM) index.
• Demonstrate its effectiveness as a general-purpose image similarity measure.
• Compare CW-SSIM performance against established similarity indices.

Main Methods:

• Developed the complex wavelet structural similarity (CW-SSIM) index based on phase changes in wavelet coefficients.
• Conducted four case studies to evaluate CW-SSIM.
• Compared CW-SSIM with Dice and Hausdorff distance metrics.

Main Results:

• CW-SSIM demonstrated superior performance in image similarity assessment compared to Dice and Hausdorff distance.
• The index proved robust to minor image rotations and translations.
• CW-SSIM provided effective comparisons without requiring image registration.

Conclusions:

• The complex wavelet structural similarity (CW-SSIM) index is a powerful and versatile tool for image similarity evaluation.
• CW-SSIM offers significant advantages in robustness, computational efficiency, and reduced preprocessing requirements.
• This new index holds promise for diverse applications in image analysis and computer vision.