Extension of phase correlation to subpixel registration

Area of Science:

• Image processing
• Computer vision
• Signal analysis

Background:

• Phase correlation is a widely used technique for image registration.
• Downsampling images can introduce artifacts that complicate traditional phase correlation analysis.
• Accurate subpixel translation estimation is crucial for various applications, including remote sensing and medical imaging.

Purpose of the Study:

• To derive analytic expressions for the phase correlation of downsampled images.
• To analyze the distribution of signal power in the phase correlation of downsampled images.
• To provide a closed-form solution for subpixel translation estimation using phase correlation of downsampled images.

Main Methods:

• Derivation of analytic expressions for phase correlation in downsampled images.
• Analysis of signal power distribution, identifying multiple coherent peaks.
• Application of derived analytic results for subpixel translation estimation and error analysis.

Main Results:

• Signal power in downsampled image phase correlation is distributed across several adjacent coherent peaks.
• These peaks are related to the polyphase transform of a filtered unit impulse.
• The analytic results offer a closed-form solution for subpixel translation estimation.
• Accurate subpixel translation estimation was achieved for various image types and spectral bands.

Conclusions:

• The derived analytic expressions accurately describe the phase correlation of downsampled images.
• The findings provide a robust method for subpixel translation estimation, overcoming downsampling challenges.
• The approach demonstrates high performance across diverse imaging scenarios and spectral ranges.