I have implemented an image based RDH algorithm using histogram shifting. I am getting a decent embedding capacity and good visual quality marked images.
As I know the three main characteristics of an RDH technique is embedding capacity, visual image quality and security or robustness. I understood the first two parameters well, but I am facing little problem in justifying robustness of my technique.
Please help me by providing some information.
Thank you.
Robustness is a subjective measurement and should be narrowed down to a specific type of attack for the quantification. For an example, we can test the reversibility against the quality factor of JPEG compression algorithm, so the result will yield how robust your algorithm against JPEG compression (because the quality factor is a quantitative measurement). If you have gone through some related literature, you would find the assessments have been done through many types of attacks, such as the reversibility vs JPEG compression quality factor, amount of added noise, cropping percentages and geometrical transformations etc.
robustness in this case can be considered as 'robustness of the hidden secret as well as the cover' against various attacks (noise, compression, transformation, cropping etc).
If the hidden message can be retrieved without significant loss/damage against the attacks as well as if the cover also can be reconstructed to its original form(since the method you claim to be reversible) then the method is robust against the attacks of study, according to me!
- Badges
- Science topic
- Similar topics
- Education