|
Concept/metric | Traditional watermark approaches | Zero-watermarking approach [1] | Zero-watermarking approach [35] | Zero-watermarking approach [34] | Zero-watermarking approach [33, 36] | Proposed watermarking method |
|
Overhead (additional) | Proportional to embedded key size | None | None | None | None | None |
|
Embedding mode | Format-encoding | Logical: zero-watermark encoding | Logical: zero-watermark encoding | Logical: zero-watermark encoding | Logical: zero-watermark encoding | Logical: zero-watermark encoding |
|
Location of watermark message () | Embedded in | Embedded in | Embedded in | Embedded in | Embedded in | Embedded in |
|
Processing and embedding decision | Based on searching through text for candidate words, lines, and spaces | Based on double-letter words in English language | Based on sentence entropy | Based on the first letter with specific word lengths | Based on comparing Unicode summations | Based on comparing Unicode summations and logical operations |
|
Compatible with various formats? | Limited | Yes, only English character support needed | Supports Chinese language only | Supports English language only | Yes, host-document language character support/Unicode needed | Yes, host-document language character support/Unicode needed |
|
Language-dependent | No | Yes | Yes | Yes | No | No |
|
Document- authenticity verification | No | Yes | Yes | Yes | Yes | Yes |
|
Tamper-detection and identification capabilities | No | No | No | Yes | No | Yes |
|
Integrity robustness | Weak | Strong | Strong | Strong | Strong | Strong |
|
Perceptibility performance | Low-medium | High | High | High | High | High |
|
Capacity ratio performance | Inversely proportional to perceptual similarity | High | High | High | High | High |
|
Capability to extract publishers watermark logo | No | Yes | No | No | No | Yes |
|