Automatic feature localisation with constrained local models

We present an efficient and robust method of locating a set of feature points in an object of interest. From a training set we construct a joint model of the appearance of each feature together with their relative positions. The model is fitted to an unseen image in an iterative manner by generating templates using the joint model and the current parameter estimates, correlating the templates with the target image to generate response images and optimising the shape parameters so as to maximise the sum of responses. The appearance model is similar to that used in the Active Appearance Models (AAM) [T.F. Cootes, G.J. Edwards, C.J. Taylor, Active appearance models, in: Proceedings of the 5th European Conference on Computer Vision 1998, vol. 2, Freiburg, Germany, 1998.]. However in our approach the appearance model is used to generate likely feature templates, instead of trying to approximate the image pixels directly. We show that when applied to a wide range of data sets, our Constrained Local Model (CLM) algorithm is more robust and more accurate than the AAM search method, which relies on the image reconstruction error to update the model parameters. We demonstrate improved localisation accuracy on photographs of human faces, magnetic resonance (MR) images of the brain and a set of dental panoramic tomograms. We also show improved tracking performance on a challenging set of in car video sequences. (C) 2008 Elsevier Ltd. All rights reserved.