Most image registration problems are formulated in an asymmetric fashion. Given a pair of images, one is implicitly or explicitly regarded as a template and warped onto the other to match as well as possible. In this paper, we focus on this seemingly arbitrary choice of the roles and reveal how it may lead to biased warp estimates in the presence of relative scaling. We present a principled way of selecting the template and explain why only the correct asymmetric form, with the potential inclusion of a blurring step, can yield an unbiased estimator. We validate our analysis in the domain of model-based face tracking. We show how the usual Active Appearance Model (AAM) formulation overlooks the asymmetry issue, causing the fitting accuracy to degrade quickly when the observed objects are smaller than their model. We formulate a novel, "resolution-aware fitting" (RAF) algorithm that respects the asymmetry and incorporates an explicit model of the blur caused by the camera's sensing elements into the fitting formulation. We compare the RAF algorithm against a state-of-the-art tracker across a variety of resolutions and AAM complexity levels. Experimental results show that RAF significantly improves the estimation accuracy of both shape and appearance parameters when fitting to low-resolution data. Recognizing and accounting for the asymmetry of image registration leads to tangible accuracy improvements in analyzing low-resolution imagery.

We propose a method to detect and identify faults in wheeled mobile robots. The idea behind the method is to use adaptive estimation to predict the outcome of several faults, and to learn them collectively as a failure pattern. Models of the system behavior under each type of fault are embedded in multiple parallel Kalman filter (KF) estimators. Each KF is tuned to a particular fault and predicts, using its embedded model, the expected values for the sensor readings. The residual, the difference between the predicted readings (based on certain assumptions for the system model and the sensor models) and the actual sensor readings, is used as an indicator of how well each filter is performing. A backpropagation neural network processes this set of residuals as a pattern and decides which fault has occurred, that is, which filter is better tuned to the correct state of the mobile robot. The technique has been implemented on a physical robot and results from experiments are discussed.