Week 5 — Copy Move Forgery Detection
Hi again, last week we mentioned what our project will aim in future, this week we made the first steps.
Scale Invariant Feature Transform (SIFT)
SIFT is a feature detection algorithm in computer vision to detect and describe local features in images. In our project SIFT will be used to detect keypoints in images which will then be compared to each other to detect copy move forgery.
In our project we will attempt to optimize results found by the paper (I. Amerini, L. Ballan, R. Caldelli, A. Del Bimbo, G. Serra. “A SIFT-based forensic method for copy-move attack detection and transformation recovery”, IEEE Transactions on Information Forensics and Security, vol. 6, issue 3, pp. 1099–1110, 2011.) which has a published MATLAB source that can be downloadable through here. We inspected their code to understand their steps, and ran it with some examples which is displayed below:
The program found multiple duplicated features which is an expected side effect of copy move forgery detection algorithms, as we can clearly see it is an image of a bookstore which makes it likely that they have some books displayed in multiple fronts.
The algorithm works like this, it first outputs the sift features to a file and then compares their uniqueness based on some given parameters, which we briefly described last week if the likelihood of any SIFT cluster is greater than the parameter it marks it as a copy move forgery. Our aim for the future is to transpile their source to python and optimize the matching system with different methods to have the highest detection possible.
