Academic
Publications
Computing Accurate Correspondences across Groups of Images

Computing Accurate Correspondences across Groups of Images,10.1109/TPAMI.2009.193,IEEE Transactions on Pattern Analysis and Machine Intelligence,Timot

Computing Accurate Correspondences across Groups of Images   (Citations: 4)
BibTex | RIS | RefWorks Download
Groupwise image registration algorithms seek to establish dense correspondences between sets of images. Typically, they involve iteratively improving the registration between each image and an evolving mean. A variety of methods have been proposed, which differ in their choice of objective function, representation of deformation field, and optimization methods. Given the complexity of the task, the final accuracy is significantly affected by the choices made for each component. Here, we present a groupwise registration algorithm which can take advantage of the statistics of both the image intensities and the range of shapes across the group to achieve accurate matching. By testing on large sets of images (in both 2D and 3D), we explore the effects of using different image representations and different statistical shape constraints. We demonstrate that careful choice of such representations can lead to significant improvements in overall performance.
Journal: IEEE Transactions on Pattern Analysis and Machine Intelligence - PAMI , vol. 32, no. 11, pp. 1994-2005, 2010
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 advocated in [5], it is preferable to compare the model, warped using the current estimate of the correspondences, with the original undeformed samples: in other words, measuring how well the model of texture “explains” the original samples...
    • ...We found that when enough features are present in the textures, no additional shape constraints are needed (see also [5], where the same observation was made, and possible options for the shape smoothness term were also discussed)...
    • ...As advocated in [5], better performance can be achieved if local brightness normalisation is applied to images (textures) and the gradient information is also incorporated as image channels...
    • ...In order to avoid a local minimum around the zero improvement hypothesis, the “current” sample is excluded from the model (lines 11‐12), as suggested in [5]...
    • ...First we perform brute-force search (as in [5]), trying several displacements within a given evaluation budget and selecting the best one...

    Kirill A. Sidorovet al. Efficient groupwise non-rigid registration of textured surfaces

    • ...Figure 5 shows the first three appearance modes of a model constructed from 300 face images of different people, with the correspondences computed automatically using the groupwise algorithm from [41]...

    Timothy F. Cootes. Deformable Object Modelling and Matching

    • ...Groupwise non-rigid registration has become an important tool in medical image interpretation [12,3,15,2,5]...
    • ...It is common to start from the affine transformation which best aligns the data [12,3,15,2,5]...
    • ...When applied to the training set itself it finds points which are shown to be an excellent initialisation for a groupwise non-rigid registration algorithm [5]...
    • ...The resulting sparse points were used to initialise a groupwise non-rigid registration algorithm [5]...

    Pei Zhanget al. Automatic Part Selection for Groupwise Registration

Sort by: