Content-based Digital Watermarking using Contrast and Directionality

Human visual system (HVS) models have been used in digital watermarking to minimize the visual effects of the watermark while increasing the strength of watermark. Such work has been applied to different watermarking schemes with varying degrees of success. Previous work at Digimarc resulted in a HVS model that inserts a high watermark signal in busy or high contrast areas, while reducing the watermark on connected directional edges where it becomes more visible. In certain instances, however, this technique inserts a high watermark signal in a region where masking due to the image is insufficient to hide the signal. For example, the watermark becomes apparent in areas with fine texture containing a dominant orientation like hair. This paper introduces a new HVS model, based on techniques that identify areas with a dominant orientation and suppress the watermark gain for those regions. Once a contrast is computed, another measurement (called directionality) is made on a small neighborhood using a standard wavelet filter set and a rotated wavelet filter set to determine if the region is highly oriented in one direction. The watermark strength gets suppressed if the corresponding area has high contrast and high directionality measure, while the gain reaches the maximum when the area has high contrast and low directionality measure. Experiments on problem images show that the proposed technique remedies the limitations of the previous HVS model to some extent, while not degrading the watermark detection performance.