Randomized Ring-Partition Fingerprinting with Dithered Lattice Vector Quantization

Image fingerprinting is required to depend on one or several keys to ensure the content security. In this paper, we propose randomized ring-partition fingerprinting with dithered lattice vector quantization which achieves the security of image hashing by randomness. First, we extract features using randomized ring-partition with a secret key. Then, we perform dithered lattice vector quantization to quantize the features. Finally, we generate a fingerprint with 90 binary bits to represent an image. Experimental results show that the proposed method achieves significant improvements in image fingerprinting in terms of robustness, discriminability, runtime, and compact compared with state-of-the-art fingerprinting ones.