Linearly estimating all parameters of affine motion using Radon transform

Area of Science:

• Computer Vision
• Image Processing
• Geometric Transformations

Background:

• Accurate motion estimation is crucial for computer vision and image processing.
• Existing projection-based methods using Radon transforms often fail to estimate all six affine parameters without iteration.
• Iterative methods for full parameter estimation are computationally expensive and sensitive to initial values.

Purpose of the Study:

• To propose a novel, direct method for estimating all six affine motion parameters using the Radon transform.
• To establish a linear model in the projection domain for affine motion estimation.
• To improve the accuracy and efficiency of motion estimation.

Main Methods:

• Utilizing the Radon transform to compute image projections.
• Studying the projection domain relationship between images with affine motion.
• Developing a direct linear model to solve for all six affine parameters.
• Employing a hierarchical framework for enhanced accuracy.

Main Results:

• Successfully estimated all six affine motion parameters directly.
• Demonstrated a linear model in the projection domain for affine motion.
• Achieved significantly better performance compared to state-of-the-art methods.
• Hierarchical framework improved result accuracy.

Conclusions:

• The proposed Radon transform-based method offers direct and accurate estimation of six affine motion parameters.
• This method overcomes the limitations of existing iterative and non-iterative techniques.
• The approach shows superior performance and potential for various computer vision applications.