Pose Parameter Estimation and Shape Distance

In this paper, a pose parameter estimation method is proposed, which provides an effective solution for the local optimal pose parameters problem in the shape distance measurements based on various level set based shape distance energy functions. To acquire invariance, explicit pose parameters representing translation, scaling and rotation are adopted in most recent level set based shape distance energy functions. However, as the optimization of these pose parameters is done through local gradient descent scheme, with the minimization of the related shape distance energy functions, the optimization could be trapped in local optimal pose parameters. To overcome this problem, the initial value of pose parameters in the local gradient scheme is set according to estimated optimal pose parameters using the estimation method proposed in this paper. Experiments shows this estimation provides an effective solution for the local optimal pose parameters problem in the minimization of shape distance energy function.