Table 6 Comparison of proposed method with some of the state-of-the-art methods
From: Robust image hashing using ring partition-PGNMF and local features
Comparison parameters | Ahmed et al. (2010) | Tang et al. (2008) | Monga and Mihcak (2007) | Khelifi and Jiang (2010) | Zhao et al. (2013) | Proposed method |
|---|---|---|---|---|---|---|
Features used | Local | Global | Local | Local | Global and local features | Global and local features |
Hash Length | 7168 bits | 320 bits | 64 floating point numbers | 250 bits | 560 bits | 848 bits |
Robust against JPEG compression | Y | Y | Y | Y | Y | Y |
Robust against rotation | NA | NA | NA | Y | Up to 5° only | All Arbitrary degree |
Ability to detect small area forgery | Y | Y | Y | NA | Y | Y |
Capability to find the counterfeit regions | Y | NA | NA | NA | Y | Y |
Optimal TPR When FPRÂ =Â 0 | 0.7843 | 0.8101 | 0.8213 | 0.8943 | 0.9157 | 0.9811 |
Optimal FPR When TPRÂ =Â 1 | 0.3265 | 0.4972 | 0.3545 | 0.6427 | 0.2187 | 0.0012 |
Average time (s) | NA | 0.9321 | 2.98 | 2.6 | 2.4 | 2.1 |