Methods and Theory : Congealing 1D Curves

Overview

Congealing is a flexible nonparametric data-driven framework for the joint alignment of data. It has been successfully applied to the joint alignment of binary images of digits, binary images of object silhouettes, grayscale MRI images, color images of cars and faces, and 3D brain volumes. This research enhances congealing to practically and effectively apply it to curve data. We develop a parameterized set of nonlinear transformations that allow us to apply congealing to this type of data.

For a summary of the research and curve alignment and classification results, please view the 4-page ICASSP paper below. Marwan's MS thesis contains a more detailed writeup and will be made available in the near future.

Sample Results


Blue curves are before alignment, red curves are after alignment and the black curve in each plot is the average of the curves.

BeforeAfter


Faculty


Collaborators


Graduate Students


Publications

  • Marwan Mattar, Michael Ross, and Erik Learned-Miller
    Nonparametric Curve Alignment.
    Intl. Conference on Acoustics, Speech and Signal Processing, 2009.
    [pdf]