Image mosaicing is a fundamental technique for creating seamless panoramic views, with applications ranging from photography to satellite mapping. This research focuses on advancing image mosaicing algorithms, particularly for high resolution and aerial images. The project evaluates existing algorithms, addresses variability in aerial data, and explores feature extraction techniques. A key contribution is the integration of Convolutional Neural Networks (CNNs) for homography estimation, enhancing mosaicing accuracy and efficiency. Additionally, feature extraction algorithms such as Scale-Invariant Feature Transform (SIFT) and Speeded Up Robust Features (SURF) are investigated. The proposed automated framework aims to scale efficiently, minimize resource consumption, and adapt to diverse aerial contexts. This paper presents a comparative analysis of image stitching algorithms, highlighting the effectiveness of CNN based homography estimation and feature extraction algorithms in improving mosaicing accuracy and seamlessness.

Authors: Bharat Mahesh, Aditya Pawar, Qusai Kader, Yashraj Deshmukh

Published in: International Conference for Internet Technology and Secured Transactions (ICITST-2024)

• Date of Conference: 4-6 November 2024
• DOI: 10.20533/ICITST.2024.0009
• ISBN: 978-1-913572-76-1
• Conference Location: St Anne’s College, Oxford University, UK