MULTIPURPOSE IMAGE HASHING BASED ON MULTISTAGE VECTOR QUANTIZATION

作者：Lv, Mei-Lei; Lu, Zhe-Ming^*; Huang, Yun-Long

来源：International Journal of Innovative Computing Information and Control, 2011, 7(12): 6865-6873.

• Perceptual hashing
• Perceptual image hash function
• Copyright protection
• Content authentication
• Multi-stage vector quantization

Perceptual image hashing has become an emerging solution for image indexing, authentication and watermarking. A perceptual image hash function maps a digital image into a fixed length binary string known as the hash value, which is invariant under changes to the image that is perceptually insignificant. In this paper, a multipurpose image hashing scheme based on Multi-Stage Vector Quantization (MSVQ) is proposed for both copyright protection and content authentication. The original gray-level image is first segmented into non-overlapping blocks. Each block is encoded by a two-stage vector quantizer to generate two indices, one for copyright protection and the other for content authentication. The obtained two index tables are then transformed into two binary images based on specific mapping functions. Finally, the authentication mark and permuted copyright logo are respectively XOR-ed with the two binary images to obtain final authentication and protection fingerprints. Compared with the existing DCTVQ based multipurpose hashing scheme, we provide two new mapping functions and our hashing process is much faster. Experimental results show the effectiveness of the proposed method.

• 出版日期2011-12
• 单位浙江大学; 衢州学院