You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi. I think your way of estimating Hue mean and variance in HSV color space is incorrect. I am only talking about HSV because I haven't read about the other two color spaces yet. Hue values live on a circle - the values in [0, 1) correspond to the circle, the values 0 and 1 are infinitely close. Your method treats them as if they are very far apart when you simply use array.mean() and array.std() to estimate the stats. This corresponds to estimating the parameters of a normal distribution. Instead, you should treat Hue as if it was sampled from a wrapped normal distribution. You can use https://en.wikipedia.org/wiki/Wrapped_normal_distribution#Estimation_of_parameters to estimate its mean and std.
The text was updated successfully, but these errors were encountered:
Hi. I think your way of estimating Hue mean and variance in HSV color space is incorrect. I am only talking about HSV because I haven't read about the other two color spaces yet. Hue values live on a circle - the values in [0, 1) correspond to the circle, the values 0 and 1 are infinitely close. Your method treats them as if they are very far apart when you simply use
array.mean()
andarray.std()
to estimate the stats. This corresponds to estimating the parameters of a normal distribution. Instead, you should treat Hue as if it was sampled from a wrapped normal distribution. You can use https://en.wikipedia.org/wiki/Wrapped_normal_distribution#Estimation_of_parameters to estimate its mean and std.The text was updated successfully, but these errors were encountered: