Fast image clustering based on compressed camera fingerprints
作者：
Highlights：
• The algorithm is composed of two stages of clustering: initial and fine clustering.
• The initial clustering stage uses compressed fingerprints which significantly reduces computational cost and memory requirements.
• The fine clustering stage uses average full fingerprints and enhances the quality of clusters.
• The computational complexity of the clustering algorithm is much less than state-of-the-art algorithms on small, medium and large datasets.
• The proposed algorithm is robust to NC≫SC problem i.e., when the number of clusters NC is much larger than size of cluster SC.
摘要
•The algorithm is composed of two stages of clustering: initial and fine clustering.•The initial clustering stage uses compressed fingerprints which significantly reduces computational cost and memory requirements.•The fine clustering stage uses average full fingerprints and enhances the quality of clusters.•The computational complexity of the clustering algorithm is much less than state-of-the-art algorithms on small, medium and large datasets.•The proposed algorithm is robust to NC≫SC problem i.e., when the number of clusters NC is much larger than size of cluster SC.
论文关键词：Image clustering,Photo response non-uniformity,Computational complexity,Source camera identification
论文评审过程：Received 29 February 2020, Revised 24 September 2020, Accepted 6 November 2020, Available online 20 November 2020, Version of Record 24 November 2020.
论文官网地址：https://doi.org/10.1016/j.image.2020.116070
