Shape Registration by Simultaneously Optimizing Representation and Transformation

Shape Registration by Simultaneously Optimizing Representation and Transformation,10.1007/978-3-540-75759-7_98,Yifeng Jiang,Jun Xie,Deqing Sun,Hung-ta

Shape Registration by Simultaneously Optimizing Representation and Transformation   (Citations: 2)
BibTex | RIS | RefWorks Download
This paper proposes a novel approach that achieves shape registration by optimizing shape representation and transformation simultaneously, which are modeled by a constrained Gaussian Mixture Model (GMM) and a regularized thin plate spline respectively. The problem is formulated within a Bayesian framework and solved by an expectation-maximum (EM) algorithm. Compared with the popular methods based on landmarks-sliding, its advantages include: (1) It can naturally deal with shapes of complex topologies and 3D dimension; (2) It is more robust against data noise; (3) The registration performance is better in terms of the generalization error of the resultant statistical shape model. These are demonstrated on both synthetic and biomedical shapes.
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...As noticed in [9, 10, 11, 12], many of these algorithms ignore the local neighborhood structure around each point, although it is important to preserve the local structure so that the objects of interest can deform only within certain physical constraints...
    • ...More research efforts [10, 11, 12] have been made to incorporate such structural restrictions, including higher-order constraints to find physically reasonable correspondence...
    • ...We tested our HMM-based matching algorithm on four benchmark shape data sets, including synthetic bumpboxes (Fig. 3A), hand profiles (Fig. 3B), contours of femurs (Fig. 3C), and silhouette profiles (Fig. 3D), which are obtained from [11, 17, 18]...

    Xiaoning Qianet al. Shape matching based on graph alignment using hidden Markov models

    • ...In the robust point matching (RPM), active shape model (ASM), active appearance model (AAM)[1], and nonrigid shape matching algorithms using the level set based contour and surface representation, shape signature harmonic embedding[3] and Gaussian mixture based shape models[4][5], the cost function is defined as...

    Xiao Donget al. Automatic Mutual Nonrigid Registration of Dense Surfaces by Graphical ...

Sort by: