Elastically Adaptive Deformable Models

Abstract

We present a novel technique for the automatic adaptation of a deformable model's elastic parameters within a Kalman filter frame-work for shape estimation applications. The novelty of the technique is that the model's elastic parameters are not constant, but time varying. The model for the elastic parameter variation depends on the local error of fit and the rate of change of the error of fit. By augmenting the state equations of an extended Kalman filter to incorporate these additional variables and take into account the noise in the data, we are able to significantly improve the quality of the shape estimation. Therefore, the model's elastic parameters are initialized always to the same value and they subsequently modified depending on the data and the noise distribution. In addition, we demonstrate how this technique can be parallelized in order to increase its efficiency. We present several experiments to demonstrate the effectiveness of our method.