Research On Adaptive Robust Reversible Watermarking Algorithm

Digital watermarking is an important part of information hiding technology,its embedded secret information is transparent to users and allow extraction of embedded watermark information from embedded digital files,while hides this information from snoopers.With the development of recent years,digital watermarking technology has many mature algorithms and is applied to real life,in which robust reversible watermark is widely used because of its unique advantages.This thesis is based from the National Natural Science Foundation project.The aim of this study is to improve the robust reversible digital watermarking algorithm,so that the new algorithm has higher robustness and lower distortion of the human eye.This research will make the application of robust reversible digital watermarking more extensive and better balance the resistance between non-malicious attack and human eye distortion.The research method of this thesis is to study the new robust reversible digital watermarking algorithm at home and abroad,improve the existing algorithm,theoretically analyze and put forward a novel algorithm,finally using MATLAB simulation and compare with the previous algorithm to ensure there is improvements over the previous algorithm.In this thesis,an adaptive robust reversible data hiding scheme is proposed by studying the relevant algorithms.When the original image is not attacked,the hidden information can be extracted completely and the original image is recovered.The embedded hidden information can still be extracted correctly even after being compressed by a certain degree due to JPEG format.This scheme divides the image into the non-overlapping blocks first,then selects the middle region in the block image to carry on the complexity forecast.According to the forecast results,different parameters were chosen to embed the blocks in two cases.This embedding has high robustness and less distortion,and follows with the observation rule of human eyes.Blocks of different complexity in embedding will change the pixels in different values according to the algorithm,this approach makes the interval between a 0-bit region and a 1-bit region different for each complex block.The extraction terminal can determine the interval so as to correctly extract the watermarking information,and the interval ensures that the embedded image can still extract the watermark correctly after the attack of JPEG compression and so on.Experimental results show that the proposed scheme has better robustness and higher subjective quality than previous schemes.In the future,the optimization of embedded parameters and dynamic extraction intervals will continue to be studied.The embedding parameters are optimized for better balance between distortion and robustness.The final extraction position of the proposed scheme is correlated with the predetermined parameter value,if the extraction interval can be selected dynamically when extracting,the dynamic selection algorithm should have higher extraction accuracy than the static selection interval.