Edge detection based JND model for digital watermarking
One of the main goals of watermarking is to optimize capacity while preserving high video fidelity. The perceptual adjustment of the watermark is mainly based on Watson Just Noticeable Difference (JND) model. Recently, it was proposed to improve Watson model using the classification blocks inherent to the encoder in a compressed stream. Although, the new model outperforms the previous one, especially in increasing the watermark power in textured blocks, it still underestimates JNDs at block's edges. This work presents a detailed comparison of these two models and proposes a new method that exploits the good characteristics of the two available models. In addition, experimental results on perceptibility are reported.