I am just wondering how can we measure how well two images "line up"? Is there a more rigorous way to explain what is "similar" and what is "different"?
qiqinl
You may view an image with m x n pixels and 3 channels as a 3mn-dimensional vector, and performing a normal dot product can tell you the answer. :)
Andy
For vectors, we have the notion of orthogonality, which corresponds to the case when the inner product is zero. Is there also a notion of orthogonality when we consider pictures as vectors? If so, what are the visual characteristics of two orthogonal pictures?
I am just wondering how can we measure how well two images "line up"? Is there a more rigorous way to explain what is "similar" and what is "different"?
You may view an image with m x n pixels and 3 channels as a 3mn-dimensional vector, and performing a normal dot product can tell you the answer. :)
For vectors, we have the notion of orthogonality, which corresponds to the case when the inner product is zero. Is there also a notion of orthogonality when we consider pictures as vectors? If so, what are the visual characteristics of two orthogonal pictures?