Image Identification Using SIFT Algorithm: Performance Analysis against Different Image Deformations

TL;DR

This paper evaluates the robustness of the SIFT algorithm in image identification tasks under various distortions like rotation, scaling, fisheye, and motion, analyzing true/false positive rates and keypoint orientation differences.

Contribution

It provides a comprehensive performance analysis of the SIFT algorithm against multiple image deformations, highlighting its strengths and limitations.

Findings

SIFT maintains high true positive rates under moderate distortions.

Performance degrades significantly with severe fisheye and motion distortions.

Distribution of keypoint orientation differences varies with the type of deformation.

Abstract

Image identification is one of the most challenging tasks in different areas of computer vision. Scale-invariant feature transform is an algorithm to detect and describe local features in images to further use them as an image matching criteria. In this paper, the performance of the SIFT matching algorithm against various image distortions such as rotation, scaling, fisheye and motion distortion are evaluated and false and true positive rates for a large number of image pairs are calculated and presented. We also evaluate the distribution of the matched keypoint orientation difference for each image deformation.