Content-Adaptive Multi-level Data Embedding for Lossless Data Hiding

Loss less data hiding, also known as reversible data hiding, is a newly developed research field and relating applications in digital rights management (DRM). Secret information can be hidden into original image at the encoder, and then image containing secret data is delivered to the decoder. For decoding, original image and secret information need to be perfectly recovered from the received image. By modifying selected pixel values in the histogram of original image, the larger amount of secret information can be embedded at the encoder. Later on, with the reasonable size of side information, both the original image and the secret information can both be recovered. We take the relationships between adjacent pixels into consideration, and propose a scheme that is able to hide more than one bit per pixel of data capacity, with the reasonable output image quality. Simulation results with our algorithm have demonstrated the better results over existing ones for real applications. With the comparisons of embedding capacity for encoding, and side information for decoding, our algorithm works well under the requirement of high data capacity for loss less data hiding.