[Paper Review] Image Matching Using SIFT, SURF, BRIEF and ORB: Performance Comparison for Distorted Images

Ebrahim Karami, Siva Prasad+1 more|arXiv (Cornell University)|Oct 7, 2017

Advanced Image and Video Retrieval Techniques8 references303 citations

TL;DR

The paper compares SIFT, SURF, BRIEF, and ORB for image matching under various distortions, evaluating robustness, keypoints, matching rate, and speed.

ABSTRACT

Fast and robust image matching is a very important task with various applications in computer vision and robotics. In this paper, we compare the performance of three different image matching techniques, i.e., SIFT, SURF, and ORB, against different kinds of transformations and deformations such as scaling, rotation, noise, fish eye distortion, and shearing. For this purpose, we manually apply different types of transformations on original images and compute the matching evaluation parameters such as the number of key points in images, the matching rate, and the execution time required for each algorithm and we will show that which algorithm is the best more robust against each kind of distortion. Index Terms-Image matching, scale invariant feature transform (SIFT), speed up robust feature (SURF), robust independent elementary features (BRIEF), oriented FAST, rotated BRIEF (ORB).

Motivation & Objective

• Motivate robust image matching for computer vision and robotics applications.
• Compare classic feature detectors/descriptors (SIFT, SURF, BRIEF, ORB) under diverse distortions.
• Identify which algorithm is most robust to each type of distortion.
• Provide practical guidance on trade-offs between accuracy and speed for distorted images.

Proposed method

• Manually apply different transformations to original images to generate distorted data.
• Compute and compare key metrics: number of keypoints detected, matching rate, and execution time for each algorithm.
• Analyze robustness of SIFT, SURF, BRIEF, and ORB against scaling, rotation, noise, fisheye distortion, and shearing.
• Evaluate which algorithm performs best under each distortion category.

Experimental results

Research questions

• RQ1 Which feature matching technique (SIFT, SURF, BRIEF, ORB) is most robust to each distortion type (scaling, rotation, noise, fisheye, shearing)?
• RQ2 How do keypoints, matching rate, and execution time vary across algorithms under different distortions?
• RQ3 What practical guidance can be given for selecting a matching method under distorted image conditions?
• RQ4 Do BRIEF and ORB offer favorable speed-accuracy trade-offs compared to SIFT and SURF under distortions?

Key findings

• SIFT, SURF, BRIEF, and ORB exhibit different robustness profiles across distortion types.
• The study reports metrics including the number of keypoints, matching rate, and execution time to compare algorithms.
• The paper identifies which algorithm is best for each distortion category based on the evaluated parameters.
• The results provide insights into trade-offs between robustness and computational efficiency for distorted image matching.