This paper describes a novel system for building seamless texture maps for a surface of arbitrary topology from real images of the object taken with a standard digital camera and uncontrolled lighting. In our application we wish to take a sparse set of real images of a 3D object, and apply the images to an approximate surface model of the object to generate a high quality textured model. In practice the measured colour and intensity for a surface element observed in different photographic images will not agree. This is due to the interaction between real world lighting effects (such as highlights and specularities) and variations in the camera gain settings as well as registration and surface modelling errors. We describe a new automatic approach that extends a classical 2D image blending technique to a 3D surface, which produces high quality photo-realistic results at a low computational cost. 1

We present a robust method for automatically matching features in images corresponding to the same physical point on an object seen from two arbitrary viewpoints. Unlike conventional stereo matching approaches we assume no prior knowledge about the relative camera positions and orientations. In fact in our application this is the information we wish to determine from the image feature matches. Features are detected in two or more images and characterised using affine texture invariants. The problem of window effects is explicitly addressed by our method-our feature characterisation is invariant to linear transformations of the image data including rotation, stretch and skew. The feature matching process is optimised for a structure-from-motion application where we wish to ignore unreliable matches at the expense of reducing the number of feature matches.

In this paper we describe an efficient method for fitting a prior linear shape model to image data using a Kalman filter framework. This work extends previous methods in several significant respects. Firstly, the dimensionality of our shape representation is varied dynamically to reflect the available information at the current search scale so that more shape parameters are used as the fitting process converges. A coarse to fine sampling strategy is used so that the computational expense of the initial few iterations is much reduced. Finally, we re-examine the aperture problem and show how the conventional use of searching along normals to the estimated curve can be improved upon. 1 Introduction The Kalman filter [1] has proven a useful tool for real-time tracking in computer vision [2, 3]. Much of this work concentrates on the problem of tracking one or more objects through a sequence of images. Blake et al [2] set out a mathematical framework for tracking contours represent...

There has been a great deal of recent interest in statistical models of 2D landmark data for generating compact deformable models of a given object. This paper extends this work to a class of parametrised shapes where there are no landmarks available. A rigorous statistical framework for the eigenshape model is introduced, which is an extension to the conventional Linear Point Distribution Model. One of the problems associated with landmark free methods is that a large degree of variability in any shape descriptor may be due to the choice of parametrisation. An automated training method is described which utilises an iterative feedback method to overcome this problem. The result is an automatically generated compact linear shape model. The model has been successfully applied to a problem of tracking the outline of a walking pedestrian in real time. 1 Introduction Statistical Analysis of 2D landmark data has become a well established tool in computer vision (e.g. morpholog...