Abstract: The conventional QIM based fingerprinting system offers relatively poor collusion resistance and image fidelity because of fixed quantization step size. For these problems, an anti-collusion fingerprinting scheme is proposed. A random signal is added to the DCT domains of host signal during embedding fingerprint. Meanwhile, the quantization step size is selected on the basis of Watson’s perceptual model. Experimental results confirm the QIM fingerprinting system based on proposed scheme has better performance on collusion resistance and perceptual quality compared with the original QIM method.

Key words: perceptual model, quantization index modulation, digital fingerprinting, performance evaluation

摘要： 在传统的基于QIM（Quantization Index Modulation）嵌入的数字指纹系统中，由于量化步长固定，指纹系统的抗共谋性能和图像保真度较差。针对这些问题，提出一种抗共谋指纹方案，在嵌入指纹时添加随机信号到宿主信号的DCT域，根据Watson视觉模型选择量化步长。实验结果表明，与原始的QIM指纹方案相比，采用提出方案的QIM指纹系统在抗共谋能力和视觉失真度方面都具有较好的性能。

关键词: 视觉模型, 量化索引调制, 数字指纹, 性能评价