Is it possible to compare two images using ANOVA? I am trying to solve a problem where there are two classes - similar image and dissimilar. I want to classify a set of test images (comparing them to a reference image) into either similar or dissimilar. Is ANOVA a good comparison tool to accomplish this task? As always, thanks in advance!
Asked
Active
Viewed 653 times
1
-
It is unclear what you mean by image similarity and how you intend to apply ANOVA to images. – Don Reba Oct 11 '16 at 09:08
-
@jd1338, you might write that as an answer; I think it's worth accepting to archive the question. – Prune Oct 11 '16 at 17:47
-
Thanks for the hint. Did that as suggested. – JSN Oct 12 '16 at 07:06
1 Answers
2
Usually, an ANOVA is used to compare mean values of more than two groups. If exactly two groups are supposed to be compared, a t-test would be used.
However, in your case it seems like you want to compare groups of n=1 several times. In such a case an ANOVA wouldn't be the method of choice. Moreover, all classic statistical methods require sample sizes > 1, which is why I doubt that your approach is very promising.
Instead, I would rather take a look at algorithms which compare images. See here for example: How can I measure the similarity between two images?
-
@troymyname00 if this answer helped, please consider selecting it as an answer by clicking on the check mark. This lets community know the answer worked and the issue is closed. – JSN Sep 10 '17 at 16:28