If "inner product", like many other operations, is arbitrary and with many lookalikes, is any operation that has those properties (positive, etc) considered also a valid "inner product"?
tiffany2
@sccao I think so, as long as it has those properties! The professor mentioned that inner products can be used to measure how "aligned" two vectors are, so I guess in different fields there could be different constraints for how that alignment is measured. Not sure what these differences could be though.
weiyuc
In terms of information, would it be better to consider the entropy or the rank of the matrix, rather than just edge?
An image of texture could be full of edges but has very few information.
Will we be covering Canny edge detection?
I always wondered how computer vision worked! wow
If "inner product", like many other operations, is arbitrary and with many lookalikes, is any operation that has those properties (positive, etc) considered also a valid "inner product"?
@sccao I think so, as long as it has those properties! The professor mentioned that inner products can be used to measure how "aligned" two vectors are, so I guess in different fields there could be different constraints for how that alignment is measured. Not sure what these differences could be though.
In terms of information, would it be better to consider the entropy or the rank of the matrix, rather than just edge? An image of texture could be full of edges but has very few information.