Local-peak scale-invariant feature transform for fast and random image stitching

TL;DR

This paper introduces LP-SIFT, a fast, scale-invariant feature detection algorithm inspired by fluid turbulence, significantly accelerating large image stitching for wide field of view applications.

Contribution

The paper presents a novel multiscale feature detection method, LP-SIFT, that enhances image stitching speed by integrating fluid turbulence-inspired local peaks with RANSAC.

Findings

Stitches nine large images within approximately 159 seconds.

Achieves order-of-magnitude speed improvement over traditional SIFT.

Demonstrates practical application in diverse fields like terrain mapping and biological analysis.

Abstract

Image stitching aims to construct a wide field of view with high spatial resolution, which cannot be achieved in a single exposure. Typically, conventional image stitching techniques, other than deep learning, require complex computation and thus computational pricy, especially for stitching large raw images. In this study, inspired by the multiscale feature of fluid turbulence, we developed a fast feature point detection algorithm named local-peak scale-invariant feature transform (LP-SIFT), based on the multiscale local peaks and scale-invariant feature transform method. By combining LP-SIFT and RANSAC in image stitching, the stitching speed can be improved by orders, compared with the original SIFT method. Nine large images (over 2600*1600 pixels), arranged randomly without prior knowledge, can be stitched within 158.94 s. The algorithm is highly practical for applications requiring a wide field of view in diverse application scenes, e.g., terrain mapping, biological analysis, and even criminal investigation.