New Multi-Domains Image Watermarking Method Based on Multi-Watermarks Embedding and Neural Network Segmentation

The need for a robust method that protects owner rights against unintended or malicious stealing, which has arisen because of the rapid increase of Internet data exchange and multimedia use, in addition to international electronic commercial services and the multimedia that it contains, has become a substantial constraint to the research in this domain. This paper proposes a novel method for image watermarking based on embedding multiple identical watermarks in different domains of the image representation: i.e. spatial and DCT domains, without any distortion of the watermarked image. In the spatial domain, the processing method uses a non-linear neural network segmentation to output the different zones of watermark embedding with respect to the image characteristics. Basing on this study, we compute the appropriate variable embedding gain factors chosen within the limit of the perceptibility threshold. However, in the DCT domain, different DCT blocks coefficients are selected with respect to the JPEG table quantization values in the middle frequency band in order to decrease the image distortions and increase the watermark robustness. Fifteen watermarks are created and embedded in these two domains in order to take advantage of both spatial domain robustness (i.e. against different asynchronous attacks), and DCT domain robustness (i.e. against jpeg compression and some other distortions). After different STIRMARK attacks, experimental results show that the proposed method is robust against several synchronous and asynchronous image attacks such as filtering, lossy compression, rotations, etc. The well-chosen locations of the different recurring embedded watermarks with variable gain factors in addition to the studied selection of the used DCT coefficients where the watermarks are coded proved that all or some watermarks 1928 Hassen Seddik et al totally survive to the specific attacks. A study detailing the robustness of the proposed algorithms and the computed distortions is also presented. The simulation results are compared with different well-known algorithms to outline the robustness and the supremacy of the proposed method.