3.配准的优化算法

Optimisation：对配准过程中的参数最优化的过程

• · Optimisation involves finding some ‘best’ paramaters according to an ‘objective function’, which is wither minimised or maximised
• · The ‘objective function’ is often related to a probability based on some model有时称为测度，表示在某一个模型中间，各个事件出现的机率

• 横轴为参数的值 纵轴为objective function 求优算法应该避免local最优值，得到全局最优

Metrics (Objective function)
1.Intra-model 同种模态
(1) Mean squared difference(minimise)
(2)Normalised cross correlation(maximise)
(3)Entropy of difference(minimise)
2.Inter-model 多模态
(1)Mutual information(maximise)
(2)Normalised mutual information(maximise)
(3)Entropy correlation coeffcient(maximise)
(4)AIR cost function(minimise)

Best Transformation Parameter
(1)Gradient Descent
(2)Regular Step Gradient Descent
(3)Conjugate Gradient
(4)Levenberg-Marquardt

Gradient Descent Optimizer