Innovative Fusion of Transformer Models with SIFT for Superior Panorama Stitching

DOI

10.2991/978-94-6463-540-9_78How to use a DOI?

Keywords

Panorama Generation; Vision Transformers; Feature Matching; Image Clustering

Abstract

In the field of image stitching, generating multiple panoramas from a large set of images is a challenging task. Traditional methods often require complex pairwise comparisons, leading to time-consuming operations that may affect accuracy and efficiency. To address this issue, this paper presents an innovative method aimed at improving computational efficiency for generating multiple panoramas in multi-class grouping. By introducing vision transformer models and cosine similarity metric, our approach enables rapid evaluation of relationships between image pairs in the initial phase, thus reducing dataset size and minimizing time-consuming feature matching operations. By initially utilizing vision transformer models to extract features from each image, we implement a cosine similarity metric to rapidly assess preliminary relationships between image pairs. This preliminary phase allows for the reduction of the dataset subjected to the more computationally intensive, Fast Library for Approximate Nearest Neighbors-based (FLANN) feature matching. Experimental results demonstrate that our method achieves a 93.34% reduction in computational time compared to traditional methods, with only an 8.12% decrease in clustering accuracy. This improvement is attributed to the effective utilization of preliminary relationship assessments to optimize the feature matching process and achieve a more efficient generation of multiple panoramas in multi-class grouping.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Zheng Xiang
PY  - 2024
DA  - 2024/10/16
TI  - Innovative Fusion of Transformer Models with SIFT for Superior Panorama Stitching
BT  - Proceedings of the 2024 2nd International Conference on Image, Algorithms and Artificial Intelligence (ICIAAI 2024)
PB  - Atlantis Press
SP  - 767
EP  - 778
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-540-9_78
DO  - 10.2991/978-94-6463-540-9_78
ID  - Xiang2024
ER  -