How to calculate the angular error
显示 更早的评论
The angular error ε is defined as the angular distance between the algorithm’s estimate of the light source (ee) and the true illuminant vector (el) in normalized
ε= Cos -1(el . ee ) this is the formula for calculation of angular error
This is the formula for calculate the angular error please give the code for that
thanking you
采纳的回答
Roger Stafford
2013-8-2
Apparently you have two 3D vectors, each of unit magnitude, el and ee, and you wish to find the angle between them. One formula would be:
e = acos(dot(el,ee));
which gives e in radians. Another formula is:
e = atan2(norm(cross(el,ee)),dot(el,ee));
which also gives e in radians. Furthermore it doesn't require that el and ee have unit magnitude. It is more accurate when e is near 0 or pi where the 'acos' function suffers larger rounding errors.
3 个评论
Jan
2013-8-2
While I agree with you, the shown formula of Fathima looks more like:
e = cos(-dot(e1, ee))
Roger Stafford
2013-8-3
编辑：Roger Stafford
2013-8-3
No, Jan, I think it is a badly copied minus one superscript and refers to the inverse of the cosine function. It is the only thing that makes sense here.
Jan
2013-8-15
Thanks, Roger. I've said already, that I agree that your suggestion make sense. But now your explanation where the minus comes from is such obvious that I cannot understand, why it was not clear to me initially.
更多回答（0 个）
类别
在 帮助中心 和 File Exchange 中查找有关 Computer Vision Toolbox 的更多信息
2013-8-2
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
