基于序贯假设校验的优化水印检测方案

数字水印在许多实际应用中,会有不同的侧重,如用之于数码相机、DVD、大型图像及多媒体数据库等情况下,自然以嵌入少和检测快为重点,这样可减少空间占用和显著提高水印检测效率,同时还可降低对宿主对象的品质影响,为此,本文提出的基于序贯假设检验的算法,以空域水印为对象,在保证水印提取检测时有足够小的虚警概率和漏警概率的前提下确定水印的最少嵌入量,并通过同条件的对比实验说明,本方法有明显的优化效果。

Watermark Optimization Based on Sequential Hypothesis Test

It is highly desirable optimizing the number of pixels(or coefficients,or fearture points) to watermark while the probability of detection errors can be controlled well and the watermark is robust to attacks .This optimization can make the watermark detection more efficient and decreases the amount of degradation to the host signal,and these are especially essential to some applications such as in large image databases,DVD,DC and video/audio data.The watermark detector in this paper is based on a sequential hypothesis test by which the optimal number of watermark pixels is determinated.Our experimental results show that with a great reduction of the total computation, about 50% savings in the number of watermark pixels are achieved comparing to the fixed sample size detector .