Table 2 Identification accuracy for manipulated images by NMF-NMF-SQ (NMF) hashing, FJLT hashing, and content-Based fingerprinting (CBF) based on FJLT and RI-FJLT hashing.
From: An Extended Image Hashing Concept: Content-Based Fingerprinting Using FJLT
Manipulations | NMF | FJLT | CBF |
|---|---|---|---|
Additive noise | Â | Â | Â |
Gaussian |
|
| 62.36% |
Salt and Pepper noise |
|
| 97.71% |
Speckle noise |
|
| 99.77% |
Blurring | Â | Â | Â |
Gaussian blurring |
|
| 99.04% |
Circular blurring |
|
| 99.09% |
Motion blurring |
|
| 99.66% |
Geometric attacks | Â | Â | Â |
Rotation |
|
| 86.54% |
Cropping |
|
| 96.14% |
Scaling |
|
| 100% |
JPEG compression |
|
| 100% |
Gamma correction |
|
| 74.26% |
the help of median filter in preprocessing, the identification accuracy of NMF hashing under Gaussian noise could be improved to 90.61% and 99.5% for FJLT hashing.
the help of median filter in preprocessing, the identification accuracy of NMF hashing under Gaussian noise could be improved to 90.61% and 99.5% for FJLT hashing.