Image description with generalized pseudo-Zernike moments.

Area of Science:

• Computer Vision
• Image Processing
• Mathematical Imaging

Background:

• Orthogonal moment functions are crucial for image representation and analysis.
• Existing methods like pseudo-Zernike and Chebyshev-Fourier moments have limitations in certain applications.
• The need for numerically stable and accurate image descriptors persists.

Purpose of the Study:

• To propose a new set of orthogonal moment functions for image description.
• To introduce generalized pseudo-Zernike polynomials for this purpose.
• To evaluate the performance of these new moments in image reconstruction and recognition tasks.

Main Methods:

• Development of orthogonal moment functions based on generalized pseudo-Zernike polynomials.
• Scaling of polynomials for enhanced numerical stability.
• Analysis of image reconstruction capability and invariant character recognition accuracy.
• Comparative experimental evaluation against pseudo-Zernike and Chebyshev-Fourier moments.

Main Results:

• The proposed generalized pseudo-Zernike moments demonstrate high performance.
• Superiority observed in both noise-free and noisy image conditions.
• Effective in both image reconstruction and invariant character recognition tasks.

Conclusions:

• Generalized pseudo-Zernike moments offer an improved approach for image description.
• The proposed method provides enhanced accuracy and stability.
• This advancement benefits various image analysis and recognition applications.