LAUSR

How to accurately register point sets still remains a challenging task, due to some unfavorable factors. In this article, a robust point set registration approach is proposed based on the Gaussian mixture model (GMM) with multiple effective constraints. The GMM is established by wrapping a model point set to a target point set, via a spatial transformation. Instead of a displacement model, the spatial transformation is decomposed as two types of transformations, an affine transformation and a nonaffine deformation. For the affine transformation, a constraint term of the parameter vector is applied to improve the robustness and efficiency. In order to enforce the smoothness, the square norm of the kernel Hilbert space is adopted as a coherent constraint for the nonaffine deformation. Moreover, the manifold regularization is utilized as a constraint in the proposed model, to capture the spatial geometry of point sets. In addition, the expectation-maximization algorithm is developed to solve the unknown variables of the proposed model. Compared to the state-of-the-art approaches, the proposed model is more robust to deformation and rotation, due to the use of multiple effective constraints. Experimental results on several widely used data sets demonstrate the effectiveness of the proposed model.