Research On Lossless Database Watermarking Algorithm

With the continuous development of big data technology,data sharing has become more common,and the amount of shared data has also increased dramatically.The popularization of big data sharing and trading,which bringing convenience to work and life,inevitably brings new adjustments to the copyright protection of big data,for example,there may be a situation of secondary distribution and secondary sales after the database distributed and sold at a lower level.Therefore,after the database leaked,how to realize the copyright protection and trace the source of the leakage has become a key issue in the research of information security.Robust database watermarking has been widely studied as an effective big data copyright protection technology,but this technology will inevitably bring distortion to the original carrier.Although the distortion is relatively small compared to the original data,in fields like medical treatment,remote sensing images,machine learning,etc.,even this subtle distortion is still unacceptable.To this end,this thesis focuses on the lossless robust reversible watermarking technology,which not only improving the security protection ability,but also ensuring the availability of the database,the specific work is as follows: 1.Propose a histogram shift algorithm for database based on grouping.By using the primary key hash algorithm to group the database,combined with the statistical characteristics of the database to filter out the best attribute column,and then calculate the difference between the group average and the original attribute item to obtain a histogram.Different from the traditional histogram translation technology,this thesis reserves a fault-tolerant space to deal with malicious attacks,which can resist common database attacks.The algorithm embeds the watermarking information into the database with low loss.After the watermarking extraction completed,the database can be restored according to the threshold and the watermarking information to achieve a lossless database watermarking.This algorithm not only recovers the carrier database non-destructively after extracting watermark,but also has high robustness when subjected to malicious attacks.2.A database zero-watermarking algorithm based on cyclic shift and a database zero watermarking algorithm based on forward and reverse sorting are proposed.Both the above algorithms first use RS coding to encode the watermarking information,and use a hash function to group the database,and "embed" the watermark into the database by changing the relative position of the group.The database zero watermark algorithm based on cyclic shift embeds the watermark into the database by cyclically moving packets,while the database zero watermark algorithm based on the forward and reverse sorting embeds the watermark into the database by changing the order of the two groups before and after.The above two algorithms do not need to modify the database attribute value,so the embedding of watermark does not affect the availability of the database,in other words,the watermarking can be embedded in the database without loss.When the database is subjected to a malicious attack,the probability density function and the centroid are used to obtain the relative position of the packet,then the embedded watermarking information can be extracted,and the final watermark is obtained by RS error correction decoding.Experiments prove that the two proposed algorithms have high robustness.