Showing papers presented at "British Machine Vision Conference in 2003"

Shape recognition with edge-based features

Abstract: In this paper we describe an approach to recognizing poorly textured objects, that may contain holes and tubular parts, in cluttered scenes under arbitrary viewing conditions. To this end we develop a number of novel components. First, we introduce a new edge-based local feature detector that is invariant to similarity transformations. The features are localized on edges and a neighbourhood is estimated in a scale invariant manner. Second, the neighbourhood descriptor computed for foreground features is not affected by background clutter, even if the feature is on an object boundary. Third, the descriptor generalizes Lowe's SIFT method to edges. An object model is learnt from a single training image. The object is then recognized in new images in a series of steps which apply progressively tighter geometric restrictions. A final contribution of this work is to allow sufficient flexibility in the geometric representation that objects in the same visual class can be recognized. Results are demonstrated for various object classes including bikes and rackets.

Interleaved Object Categorization and Segmentation

Abstract: Historically, figure-ground segmentation has been seen as an important and even necessary precursor for object recognition In that context, segmentation is mostly defined as a data driven, that is bottom-up, process As for humans object recognition and segmentation are heavily intertwined processes, it has been argued that top-down knowledge from object recognition can and should be used for guiding the segmentation process In this paper, we present a method for the categorization of unfamiliar objects in difficult real-world scenes The method generates object hypotheses without prior segmentation that can be used to obtain a category-specific figure-ground segmentation In particular, the proposed approach uses a probabilistic formulation to incorporate knowledge about the recognized category as well as the supporting information in the image to segment the object from the background This segmentation can then be used for hypothesis verification, to further improve recognition performance Experimental results show the capacity of the approach to categorize and segment object categories as diverse as cars and cows

Exact Polyhedral Visual Hulls

Abstract: We propose an exact method for efficiently and robustly computing the visual hull of an object from image contours. Unlike most existing approaches, ours computes an exact description of the visual hull polyhedron associated to polygonal image contours. Furthermore, the proposed approach is fast and allows real-time recovery of both manifold and watertight visual hull polyhedra. The process involves three main steps. First, a coarse geometrical approximation of the visual hull is computed by retrieving its viewing edges, an unconnected subset of the wanted mesh. Then, local orientation and connectivity rules are used to walk along the relevant viewing cone intersection boundaries, so as to iteratively generate the missing surface points and connections. A final connection walkthrough allows us to identify the planar contours for each face of the polyhedron. Implementation details and results with synthetic and real data are presented.

Facial Feature Detection using AdaBoost with Shape Constraints

Abstract: Recently a fast and efficient face detection method has been devised [11], which relies on the AdaBoost algorithm and a set of Haar Wavelet like features. A natural extension of this approach is to use the same technique to locate individual features within the face region. However, we find that there is insufficient local structure to reliably locate each feature in every image, and thus local models can give many false positive responses. We demonstrate that the performance of such feature detectors can be significantly improved by using global shape constraints. We describe an algorithm capable of accurately and reliably detecting facial features and present quantitative results on both high and low resolution image sets.

Adding Curvature to Minimum Description Length Shape Models.

Abstract: The Minimum Description Length (MDL)approach to shape modelling seeks a compact description of a set of shapes in terms of the coordinates of marks on the shapes. It has been shown that the mark positions resulting from this optimisation to a large extent solve the so-called point correspondence problem: How to select points on shapes defined as curves so that the points correspond across a data set. However, this MDL approach does not capture important shape characteristics related to the curvature of the curves, and occasionally it places marks in obvious conflict with the human notion of point correspondence. This paper shows how the MDL approach can be fine-tuned by adding a term to the cost function expressing the mismatch of curvature features across the data set. The method is illustrated on silhouettes of adult heads. The MDL method is able to solve the point correspondence problem and a classification of the heads into male and female improves dramatically when using the MDL-generated marks.

Synthesis of Photographic Quality Facial Composites using Evolutionary Algorithms

Abstract: A facial composite system is described for use in criminal investigations which has distinct advantages over current methods. Unlike traditional feature based methods, our approach uses both local and global facial models, allowing a witness to evolve plausible, photo-realistic face images in an intuitive way. The basic method combines random sampling from a facial appearance model (AM) with an evolutionary algorithm (EA) to drive the search procedure to convergence. Three variants of the evolutionary algorithm have been explored and their performance measured using a computer simulation of a human witness (virtual witness). Further system functionality, provided by local appearance models and transformations of the appearance space which respectively allow both local features and semantic facial attributes to be manipulated, is presented. Preliminary examples of composites generated with our system are presented which demonstrate the potential superiority of the evolutionary approach to composite generation.

Synchronizing Image Sequences of Non-Rigid Objects.

Abstract: For stereopsis, images of a given scene must be captured at the same instant to ensure temporal consistency. For sequences of images (i.e. video streams) this requires the potentially costly and technically complex process of synchronizing cameras. We present a simple but effective method for automatically recovering the sub-frame temporal offset between image sequences taken using unsynchronized cameras. Having recovered the offset, we obtain the affine structure of a non-rigid motion. The technique is demonstrated for the application of human motion capture.

Minimizing the description length using steepest descent

Abstract: Recently there has been much attention to MDL and its effectiveness in automatic shape modelling. One problem of this technique has been the slow convergence of the optimization step. In this paper the Jacobian of the objective function is derived. Being able to calculate the Jacobian, a variety of optimisation techniques can be considered. In this paper we apply steepest descent and show that it is more efcient than the previously proposed Nelder-Mead Simplex optimisation.

A continuous Formulation of intrinsic Dimension

Abstract: The intrinsic dimension (see, e.g., [29, 11]) has proven to be a suitable descriptor to distinguish between different kind of image structures such as edges, junctions or homogeneous image patches. In this paper, we will show that the intrinsic dimension is spanned by two axes: one axis represents the variance of the spectral energy and one represents the a weighted variance in orientation. Moreover, we will show in section that the topological structure of instrinsic dimension has the form of a triangle. We will review diverse definitions of intrinsic dimension and we will show that they can be subsumed within the above mentioned scheme. We will then give a concrete continous definition of intrinsic dimension that realizes its triangular structure.

Learning a Kinematic Prior for Tree-Based Filtering

Abstract: The aim in this paper is to track articulated hand motion from monocular video. Bayesian filtering is implemented by using a tree-based representation of the posterior distribution. Each tree node corresponds to a partition of the state space with piecewise constant density. In a hierarchical search regions with low probability mass can be rapidly discarded, while the modes of the posterior can be approximated to high precision. Large sets of training data are captured using a data glove, and two techniques for constructing the tree are described: One method is to cluster the collected data points using a hierarchical clustering algorithm, and use the cluster centres as nodes. Alternatively, a lower dimensional eigenspace can be partitioned using a grid at multiple resolutions, and each partition centre corresponds to a node in the tree. The effectiveness of these techniques is demonstrated by using them for tracking 3D articulated hand motion in front of a cluttered background.

Linking Sequences of Active Appearance Sub-Models via Constraints: an Application in Automated Vertebral Morphometry

Abstract: Statistical models of shape and appearance are powerful tools for interpreting medical and other images. However there can remain problems with under-trained models being too constrained. We have combined a global model with a sequence of partially overlapping sub-models, in a manner that exploits all the statistical information, whilst mitigating the under-training problem. Instead of applying one global model, we use a global model to apply iteratively-updated soft constraints on a sequence of sub-models. These sub-models may also partially overlap, and thus previously fit sub-models can also impose soft constraints on the next iteration. The algorithm has been applied to dual x-ray absorptiometry scans of the spine in order to automate vertebral morphometry measurements, using overlapping triplets of vertebrae as the sub-models, together with a global model of the entire spine. Combining a global model in this way with a sequence of sub-models gives substantially better results than using the former alone.

Simultaneous Pose Estimation of Multiple People using Multiple-View Cues with Hierarchical Sampling

Abstract: We present a novel method for dynamic estimation of pose of multiple people using multiple video cameras. Tracking is performed using a model-based approach and a set of cues which exploit both shape and colour information. For shape we propose a fast line-search method to incorporate multi-view constraints without the computational overhead of a voxel representation. The tracking algorithm is a new hierarchical stochastic sampling scheme. Results are presented using natural movements of up to two people sharing the same capture volume. Tracking is shown to be robust over a range of natural movements, including considerable occlusion. Processing times for tracking two people are at least as short as those for tracking one person using other stochastic schemes. The high performance and efficiency are attributed to the hierarchical search method and the accuracy of the cues in identifying suitable poses.

Active Wavelet Networks for Face Alignment.

Abstract: The active appearance model (AAM) algorithm has proved to be a successful method for face alignment and synthesis. By elegantly combining both shape and texture models, AAM allows fast and robust deformable image matching. However, the method is sensitive to partial occlusions and illumination changes. In such cases, the PCA-based texture model causes the reconstruction error to be globally spread over the image. In this paper, we propose a new method for face alignment called active wavelet networks (AWN), which replaces the AAM texture model by a wavelet network representation. Since we consider spatially localized wavelets for modeling texture, our method shows more robustness against partial occlusions and some illumination changes.

Stratified Dense Matching for Stereopsis in Complex Scenes.

Abstract: Local joint image modeling in stereo matching brings more discriminable and stable matching features. Such features reduce the need for strong prior models (continuity) and thus algorithms that are less prone to false positive artefacts in general complex scenes can be applied. One of the principal quality factors in area-based dense stereo is the matching window shape. As it cannot be selected without having any initial matching hypothesis we propose a stratified matching approach. The window adapts to high-correlation structures in disparity space found in pre-matching which is then followed by final matching. In a rigorous ground-truth experiment we show that Stratified Dense Matching is able to increase matching density 3×, matching accuracy 1.8×, and occlusion boundary detection 2× as compared to a fixed-size rectangular windows algorithm. Performance on real outdoor complex scenes is also evaluated.

Dense matching using correlation: new measures that are robust near occlusions

Abstract: In the context of computer vision, matching can be done using correlation measures. This paper presents the classification of fifty measures into five families. In addition, eighteen new measures based on robust statistics are presented to deal with the problem of occlusions. An evaluation protocol is proposed and the results show that robust measures (one of the five families), including the new measures, give the best results near occlusions.

Automated Registration of 3D Faces using Dense Surface Models

Abstract: Dense surface models can be used to register unseen surfaces, using an algorithm which is a hybrid of iterative closest-point (ICP) and active shape model (ASM) fitting. In this paper we give details of this procedure and show how it can be improved by sequentially extending the transform group over which it operates. We also evaluate it for robustness to the position of the target and to shape variation across a set of unseen examples. The fit was successful on all 21 examples in our test set, with an average RMS error of 3.0mm. An initial comparison of 3 people landmarking the same scans suggests that this is within the normal landmark reproducibility range for 3D face scans.

Classification of Moving Targets Based on Motion and Appearance

Abstract: We describe a system for detection and classification of moving targets. The system includes change detection and tracking modules, which are based on adaptive background, updated with information from target level processing. The classification module performs hybrid classification which combines motion and appearance features. The system is able to perform real time detection, tracking and classification of different types of targets in natural, real life setting. Experiments demonstrate that the proposed hybrid architecture of classifiers improves classification significantly. In this paper we present a surveillance system that is able to detect, track and classify moving targets. Our system extracts static and dynamic characteristic features of moving targets and uses them to assign the targets to one of several predefined categories. The system requires minimal user input, and is able to work under diverse illumination conditions, including noisy background, while using various types of video sensors. The system provides real-time video performance due to short detection, lock-on and classification time for each target. We can logically divide the system into three major parts: the target detection module, the target tracking module and the classification module. These modules were designed to support classification of the following categories: vehicle, animal, human, group of people, crawling man and others. The system supports classification of activities and complex motions (human with a carriage, human on a bicycle). To provide reliable, view independent, classification we combine features based on motion characteristics with features based on target’s shape. By so doing we create a hybrid system that uses shape and motion for classification. Reliable classification strongly depends on the quality of objects segmentation, so significant effort was put into improving targets segmentation and tracking. In subsequent sections we describe individual modules of the system. In Section 2 we briefly describe the target detection and target tracking modules. Detailed description of these modules can be found in [14]. In Section 3 we give a de

Using Local Context To Improve Face Detection

Abstract: Most face detection algorithms locate faces by classifying the content of a detection window iterating over all positions and scales of the input image. Recent developments have accelerated this process up to real-time performance at high levels of accuracy. However, even the best of today’s computational systems are far from being able to compete with the detection capabilities of the human visual system. Psychophysical experiments have shown the importance of local context in the face detection process. In this paper we investigate the role of local context for face detection algorithms. In experiments on two large data sets we find that using local context can significantly increase the number of correct detections, particularly in low resolution cases, uncommon poses or individual appearances as well as occlusions.

Learning Based Resolution Enhancement of Iris Images

Abstract: Iris recognition is one of the most reliable personal identification methods. The potential requirement of obtaining high accuracy is that users supply iris images with good quality. It is thus necessary for an iris recognition system to operate the possibly blurred iris images due to less cooperation of users and camera with low resolution. This paper proposes a new algorithm for resolution enhancement of iris images captured by the low resolution camera in less cooperative situations. The prior probability relation between the information of different frequency bands of iris features useful for recognition is firstly learned. Then, it is incorporated into resolution enhancement algorithms to recover the lost information for the seriously blurred images. A large number of experiments on the CASIA iris database demonstrate the validity of the proposed approach.

3D head tracking using non-linear optimization.

Abstract: Accurate and reliable tracking of the 3D position of human heads is a continuing research problem in computer vision. This paper addresses the specific problem of model-based tracking with a generic deformable 3D head model. Following the work of Vetter and Blanz, a collection of head models is obtained from a 3D scanner, registered and parameterized to give a generic head model which is linearly parameterized by a small number of parameters. This is the 3D analogue of Cootes and Taylor’s active appearance models. We cast tracking as a parameter estimation problem, and note that many existing solutions to the problem—such as CONDENSATION and Kalman filtering—are analogous to nonlinear optimization strategies in numerical analysis. We show how careful analysis of the error function, parameterization of the model pose parameters, and choice of optimizer allows us to robustly track 3D head pose in digital video camera footage of quickly moving heads.

Tracking in a Hough Space with the Extended Kalman Filter.

Abstract: A combined tracking method using the Kalman filter and Hough transform is presented. An extended Kalman filter is used to model the parameters and motion of a set of lines detected in a Hough space The integration of these two techniques gives a number of advantages. The use of a Hough transform provides resilience to noise and partial occlusion, and the Kalman filter’s ability to predict future states is used to reduce the computational load of line detection. Analysis of the tracker from synthetic data shows that it is robust to noise, occlusion, and deviations from the constant motion model underlying the Kalman filter. Tracking results from video sequences illustrate its applicability to real-world domains.

The Art of Scale-Space.

Abstract: Artists pictures rarely have photo-realistic detail. Tools to create pictures from digital photographs might, therefore, include methods for removing detail. These tools such as Gaussian and anisotropic diffusion filters and connected-set morphological filters (sieves) remove detail whilst maintaining scale-space causality, in other words new detail is not created using these operators. Non-photorealistic rendering is, therefore, a potential application of these vision techniques. Sieves, in particular, preserve the appropriate edges of retained segments of interest. The resulting image has fewer extrema and is perceptually simpler than the original and is a step towards an artistic nonphotorealistic rendering of the origina. By increasing the amount of simpli- fication towards the margins of the image, the picture composition can be modulated to direct attention to centre of interest of the image. A second artistic goal. The process also removes that detail that provides perceptual cues about texture. This allows the 'eye' to readily accept alternative, artistic, textures introduced to further create an artistic impression. Moreover, the edges bounding segments accurately represent shapes in the original image and so provide a starting point for sketches.

Bayesian Stochastic Mesh Optimization for 3D reconstruction.

Abstract: We describe a mesh based approach to the problem of structure from motion The input to the algorithm is a small set of images, sparse noisy feature correspondences (such as those provided by a Harris corner detector and cross correlation) and the camera geometry plus calibration The output is a 3D mesh, that when projected onto each view, is visually consistent with the images There are two contributions in this paper The first is a Bayesian formulation in which simplicity and smoothness assumptions are encoded in the prior distribution The resulting posterior is optimized by simulated annealing The second and more important contribution is a way to make this optimization scheme more efficient Generic simulated annealing has been long studied in computer vision and is thought to be highly inefficient This is often because the proposal distribution searches regions of space which are far from the modes In order to improve the performance of simulated annealing it has long been acknowledged that choice of the correct proposal distribution is of paramount importance to convergence Taking inspiration from RANSAC andimportance sampling we craft a proposal distribution that is tailored to the problem of structure from motion This makes our approach particularly robust to noise and ambiguity We show results for an artificial object and an architectural scene

Discriminant feature selection for texture classification

Abstract: The computational complexity of a texture classification algorithm is limited by the dimensionality of the feature space. Although finding the optimal feature subset is a NP-hard problem [Boz, 2002], a feature selection algorithm that can reduce the dimensionality of problem is often desirable. In this paper, we report work on a feature selection algorithm for texture classification using two subband filtering methods: a full wavelet packet decomposition and a Gabor type decomposition. The value of a cost function associated with a subband (feature) is used as a measure of relevance of that subband for classification purposes. This leads to a fast feature selection algorithm which ranks the features according to their measure of relevance. Experiments on a range of test images and both filtering methods provide results that are promising.

Texture Edge Detection Using the Compass Operator

Abstract: The compass operator has proven to be a useful tool for the detection of color edges in real images. Its fundamental contribution is the comparison of oriented distributions of image features over a local area at each pixel. This paper presents extensions and modifications to the operator to make it applicable to texture edge detection in high dimensional images whose dimensions represent the output of a texture filter bank. The results show that the extended compass operator can robustly locate edges in natural scenes with complex textures. In addition, the use of a dynamic time warping distribution matching metric and jittered application of the operator improves the computational running time by a factor of over 50 while still producing comparable results. This large-scale speedup makes application of the algorithm to an entire image database computationally feasible.

A Multiple View Layered Representation for Dynamic Novel View Synthesis.

Abstract: We propose a multiple view layered representation for tracking and segmentation of multiple objects in a scene. Existing layered approaches are dominated by the single view case and generally exploit only motion cues. We extend this to integrate static, dynamic and structural cues over a pair of views. The goal is to update coherent correspondence information sequentially, producing a multi-object tracker as a natural byproduct. We formulate a MAP solution for estimating layer parameters which are consistent across views, with the EM algorithm used to determine both the hidden segmentation labelling and motion parameters. A persistent representation of occupancy is maintained in spite of occlusion without enforcing a particular parametric shape model. An immediate application is dynamic novel view synthesis, for which our layered approach offers a direct and convenient representation.

Video Indexing using Motion Estimation

Abstract: Summarising video data is essential to enable content-based video indexing and retrieval. A novel graph theoretic approach is presented to extract representative key frames corresponding to the shortest path of the graph for each shot. We distinguish further amongst paths of similar weight by examining the standard deviation of their constituent edge weights which improves the distribution of the selected key frames. The perceived camera motions contained within each shot are also annotated to introduce a further level of indexing and searching video content.

Rapid rendering of apparent contours of implicit surfaces for real-time tracking

Abstract: This paper addresses the problem of real-time visual tracking of structures with curved surfaces by localising the apparent contour in each frame of an image sequence. A scheme is presented for rapidly rendering the apparent contour from a predicted pose. Errors between this contour and the observed contour are then used to update the pose estimate for tracking. The rendering algorithm makes use of two contributions. Firstly, a differential equation is derived which traces out the contour generators on an iso-surface of a scalar field. Secondly a set of rules for determining the visibility of each part of the apparent contour is presented. These techniques are used to render structures of moderate complexity in under 30ms which permits real-time tracking at video frame rate.