Matching of Images with Rotation Transformation Based on the Virtual Electromagnetic Interaction

TL;DR

This paper introduces a novel image matching method for rotation transformations inspired by electromagnetic interactions, utilizing virtual currents derived from image edges to improve registration accuracy.

Contribution

The paper proposes a new approach that models image features as virtual currents and uses electromagnetic force concepts for rotation-invariant image matching.

Findings

Effective matching of rotated images demonstrated

Virtual electromagnetic force improves registration accuracy

Potential for application in image registration tasks

Abstract

A novel approach of image matching for rotating transformation is presented and studied. The approach is inspired by electromagnetic interaction force between physical currents. The virtual current in images is proposed based on the significant edge lines extracted as the fundamental structural feature of images. The virtual electromagnetic force and the corresponding moment is studied between two images after the extraction of the virtual currents in the images. Then image matching for rotating transformation is implemented by exploiting the interaction between the virtual currents in the two images to be matched. The experimental results prove the effectiveness of the novel idea, which indicates the promising application of the proposed method in image registration.