摘要
In this paper, we use a grouping method and feature marking to ensure a good localization accuracy. Then, on the basis of vertex insertion and an embedding strategy that can resist rotation, translation and scaling (RST) attacks, we design a group correlation method. And the RST invariant authentication information are generated with the log-polar transformation (LPT) and encryption technology. We construct two datasets and the invisibility, validity and localization ability are demonstrated through experiments. It can resist the batch features deletion attack accurately and mark out the deleted group.
- 出版日期2018
- 单位哈尔滨工程大学