Showing papers in "Image and Vision Computing in 1994"

TL;DR: This paper describes a technique for building compact models of the shape and appearance of flexible objects seen in 2D images, derived from the statistics of labelled images containing examples of the objects.

TL;DR: It is described how two-dimensional face images can be converted into one-dimensional sequences to allow similar techniques to be applied and how a HMM can be used to automatically segment face images and extract features that can be use for identification.

TL;DR: The algorithm is very fast, taking 0.2 seconds on a Sun SparcStation-1 for a 128 × 128 image, and is purely local and highly parallelizable (parallel implementation included).

TL;DR: A more flexible vision-based approach, which can estimate the direction of gaze from a single, monocular view of a face, which makes minimal assumptions about the structure of the face, requires lew image measurements, and produces an accurate estimate of the facial orientation.

TL;DR: A system that combines stereo vision with a 5-DOF robotic manipulator to enable it to locate and reach for objects in an unstructured environment using an affine stereo algorithm, a simple but robust approximation to the geometry of stereo vision.

TL;DR: This paper discusses some interesting properties of the Gaussian derivative kernels, like their orthogonality and behaviour with noisy input data, and describes the computer implementation of the calculation of higher order invariant structure.

TL;DR: Some of the basic concepts of the Multiscale Medial Axis are presented, ils computation is described, and some preliminary results of its application to medical images from variety of imaging modalities are demonstrated.

TL;DR: Three problems related to the analysis of facial images are addtessed: the illuminant direction, the compensation of illumination effects and, finally, the recovery of the pose of the face, restricted to in-depth rotations are proposed.

TL;DR: Some strategies for tracking with emphasis on practical importance include beam search for resolving multiple matches, support of existence for discarding false matches, and locking on reliable tokens and maximizing local rigidity for handling combinatorial explosion are introduced.

TL;DR: A realtime active surveillance vision system which detects moving objects in an everyday environment, directs the gaze of a headkye platform towards the objects and subsequently pursues them smoothly, and can continue over extended periods.

TL;DR: This paper proposes an efficient noniterative algorithm to determine the coordinates of the end points, the length, and the normal parameters of a straight line using the Hough transform, based on an analysis of the spread of votes in the accumulator array cells.

TL;DR: A new approach is described for reconstructing coronary arteries from two sequences of projection images based on a ‘predictionprojection-optimization’ loop that copes with time varying properties, deformations and superpositions of vessels.

TL;DR: The algorithm is developed and experiments are performed, proving its superior performance in terms of estimate accuracy, convergence, robustness and better segmentation.

TL;DR: The new method, called Stretch-Correlation, is based on the cross-correlation of warped image blocks which have been preprocessed to maximize the useful information content and effectively relaxes the front-o-parallel constraint which is normally imposed in area-based disparity calculation.

TL;DR: A novel algorithm is presented which makes effective use of the ground-plane constraint to derive pose estimates, and a form of the generalized Hough transform is used to group evidence from line features, and to identify approximate poses.

TL;DR: An automatic system using dynamic image sequence processing to count people getting in and out of a bus is presented, which has the main advantages are the execution speed and the reliability of the counting process.

TL;DR: The benefit of applying features at multiple scales, as well as the effects of first- and second-order information on the results are evaluated.

TL;DR: An algorithm is described that can track edges determined by the Laplacian of Gaussian operator from low to high resolution by using the image content to find the topology of the scale-space image.

TL;DR: A new method of auto-calibration for a visual sensor mounted on a robotic head based on the tracking of stationary targets, while the robotic system performs a specific controlled displacement, namely a fixed axis rotation is proposed.

TL;DR: A new algorithm based on surface fitting that makes use of anatomical knowledge of adjacency of identified anatomical structures to solve the 3D rigid body registration problem is presented.

TL;DR: This work addresses the problem of identifying specific objects within three-dimensional data sets by using a controlled, deformable model, based on superquadric geometric primitives as an initial estimate and applying genetic algorithms as the technique for solving the complex optimization problem of defining an approximate encompassing envelope within which the object will be found.

TL;DR: This work presents a new paradigm for shape from shading based upon a scale space surface representation, which becomes one of forming a surface from Gaussian basis functions in such a way that the shading in the original image is explained.

TL;DR: It is shown that a popular and apparently rigorous approach to testing, where the choice is between a fixed number of faces, each of which appears in each of afixed number of different conditions, can produce unrepresentative results.

TL;DR: Slow cross-dissolves between one digital image and another can be controlled very accurately, but as with standard film techniques, considerable ghosting of the two end-stop frames is visible during transformation.

TL;DR: After reviewing the state-of-the-art in the field of vector quantization, this work focuses on iterative and non-iterative codebook generation algorithms.

TL;DR: Using this unified approach to solve FAMIS physical model, a priori knowledge about functions and images can be jointly taken into account to improve the estimation of the underlying structures.

TL;DR: It is shown mathematically and experimentally that an index based on hierarchical features is more computationally efficient than one based on non-hierarchical features under some very general conditions which are easily satisfied.

TL;DR: A method similar to assumption-based truth maintenance systems for the collating and reasoning processes required in the labelling of input images for the recognition of objects from medical images is proposed.

TL;DR: This paper describes I-BOL-an application-specific high level programming language intended for implementing low-level image processing applications on parallel architectures, designed to be capable of implementation on distributed memory parallel machines such as transputer networks.