Abstract:
Automatic image registration, which aligns several images acquired from different viewpoints or timeframes, plays an important role in the analysis and preservation of cu...Show MoreMetadata
Abstract:
Automatic image registration, which aligns several images acquired from different viewpoints or timeframes, plays an important role in the analysis and preservation of cultural heritage. It serves as a crucial preprocessing step for performing comparative analysis, restoration, and reconstruction of cultural heritage images. In this paper, an automatic image registration approach is proposed that combines the strengths of both the Scale Invariant Feature Transform (SIFT) and the Accelerated KAZE (AKAZE) for accurate and robust registration of heritage images. While SIFT excels in matching images with varying scales and rotations, AKAZE is renowned for its robustness to non-linear transformations and challenging imaging conditions. This fusion of SIFT and AKAZE feature detection leads to a substantial increase in accurate correspondences, enhancing registration results. The effectiveness of this proposed approach has been assessed using the Affine Covariant Regions Dataset and the Indian Heritage Image Retrieval Dataset (IHIRD). The outcomes of the experiments show that the proposed method increases the number of correct correspondences when compared to those of SIFT or AKAZE alone.
Date of Conference: 13-15 December 2023
Date Added to IEEE Xplore: 19 February 2024
ISBN Information: