This question in the relation to calculating the horizontal, vertical, and diagonal correlation of a cipher image in an encryption algorithm. I have read that the correlation values should be as low as possible. What if the value is negative? Should it be closer to zero or as negative as possible?
If I have
Cipher Image 1 = diagonal correlation of -0.03204 and
Cipher Image 2 = diagonal correlation of -0.000333
Which correlation value is better?
Related to your query, I suggest you to follow https://www.researchgate.net/post/What_is_the_minimum_value_of_correlation_coefficient_to_prove_the_existence_of_the_accepted_relationship_between_scores_of_two_of_more_tests
https://www.sciencedirect.com/topics/mathematics/pearson-correlation-coefficient
Encrypted outputs should look random. So the correlation between adjacent pixels should be low and close to zero as possible
CC & NCC are two important objective performance metrics to measure the correlation between two images. The correlation coefficient (CC) between two images lies in between -1 to 1. But normalized correlation coefficient (NCC) lies in between 0 to1.
In the case of CC -1 denotes the two images are totally different, and 1 denotes two images are the same.
Dear Priyansi
In an encrypted image Pixels must be statistically independent, so that knowledge of the value of a Pixel does not provide information about the value of others.
There are different metrics for measuring that correlation.
Pearson's linear correlation coefficient only measures linear correlation, this coefficient can be zero(ideal). To detect nonlinear correlation you should use other metrics, such as Mutual Information Based on Shannon's entropy.
- Badges
- Science topic
- Similar topics
- Computer Science
- Images