Differential Energy Watermarking Based on Macro-block Prediction

Differential energy watermarking considers the properties that adjacent image blocks have similar energy in the transform domain and it is not sensitive for human system to the characteristics of high frequency, by adjusting the transform domain coefficients to achieve the purpose of watermark. This paper presents a differential energy watermarking method based on macro-block prediction. This method derived from the macro-block in the current coding frame and its best match-block in the reference frame are similar in energy, by adjusting DCT coefficients to produce the energy difference between the macro-block to achieve the purpose of watermark. The modification of DCT coefficients are within the DCT coefficients´ just noticeable difference (JND) which is multiplied by the base and the luminance adaptation effect, the contrast masking effect, and the temporal modulation factor. So the watermarking is invisibility in spatial and temporal domain. Because the watermark is embedded into DCT domain, so we need not decompress the video while embedding and extraction, it has good real-time property. Experiments show that the watermarking method can resist the noise jamming, filter and video recompression attacks.