A Digital Watermarking Scheme Based On Texture Measures In Wavelet Domain

JND threshold is usually used to control the embedding location and intensity in image watermarking system, and to determine the extraction location in blind algorithm as well. A novel JND computation method based on the low frequency coefficients of wavelet domain and texture measures is present to enhance the robustness of JND threshold. Most of the embedded watermarks can be effectively located even when the image is largely altered, thus it can improve efficaciously the detection result in blind extraction. Besides, a novel blind detection algorithm called matching detection based on statistical characteristics is given. Matching detection enhances detecting responses and obtains high credibility considering the probability of false alarm.