Reproduce “Alignment of Similarity-Transformed Images Based on Fourier--Mellin Transform Using Auxiliary Function Method” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shinji Yamashita, Yuma Kinoshita, Hitoshi Kiya
An enhanced Fourier-Mellin transform-based method for image alignment integrates a subpixel-accurate correlation maximization algorithm, based on the auxiliary function method, into the conventional Fourier-Mellin framework. This approach reduces estimation errors for translation, scale, and rotation parameters by up to 83% for translation and 80% for rotation compared to a baseline discrete cross-correlation method.
Key insights
- Replacing discrete cross-correlation maximization with an iterative, continuous optimization approach (auxiliary function method) within the Fourier-Mellin framework significantly improves subpixel alignment accuracy for similarity transformations.
- The use of log-magnitude spectra and mean subtraction before correlation in the log-polar domain is crucial for mitigating bias from low-frequency components, thereby improving the robustness and accuracy of scale and rotation estimation.
- Achieving enhanced subpixel accuracy through iterative optimization introduces a trade-off, resulting in a 2.75 to 3.08 times increase in computational execution time compared to methods relying solely on discrete cross-correlation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.