Horst Wildenauer

Detection and matching of rectilinear structures

Indoor and outdoor urban environments posses many regularities which can be efficiently exploited and used for general image parsing tasks. We present a novel approach for detecting rectilinear structures and demonstrate their use for wide baseline stereo matching, planar 3D reconstruction, and computation of geometric context. Assuming a presence of dominant orthogonal vanishing directions, we proceed by formulating the detection of the rectilinear structures as a labeling problem on detected line segments. The line segment labels, respecting the proposed grammar rules, are established as the MAP assignment of the corresponding MRF. The proposed framework allows to detect both full as well as partial rectangles, rectangle-in-rectangle structures, and rectangles sharing edges. The use of detected rectangles is demonstrated in the context of difficult wide baseline matching tasks in the presence of repetitive structures and large appearance changes.

Illumination insensitive eigenspaces

Variations in illumination can have a dramatic effect on the appearance of an object in an image. In this paper we propose how to deal with illumination variations in eigenspace methods. We demonstrate that the eigenimages obtained by a training set under a single illumination condition (ambient light) can be used for recognition of objects taken under different illumination conditions. The major idea is to incorporate a set of gradient based filter banks into the eigenspace recognition framework. This can be achieved since the eigenimage coefficients are invariant for linearly filtered images (input and eigenimages). To achieve further illumination insensitivity we devised a robust procedure for coefficient recovery. The proposed approach has been extensively evaluated on a set of 2160 images and the results were compared to other approaches.

Evaluation of Motion Segmentation Quality for Aircraft Activity Surveillance

Recent interest has been shown in performance evaluation of visual surveillance systems. The main purpose of performance evaluation on computer vision systems is the statistical testing and tuning in order to improve algorithms reliability and robustness. In this paper we investigate the use of empirical discrepancy metrics for quantitative analysis of motion segmentation algorithms. We are concerned with the case of visual surveillance on an airports apron, that is the area where aircrafts are parked and serviced by specialized ground support vehicles. Robust detection of individuals and vehicles is of major concern for the purpose of tracking objects and understanding the scene. In this paper, different discrepancy metrics for motion segmentation evaluation are illustrated and used to assess the performance of three motion segmentors on video sequences of an airports apron.

Vanishing Point Detection in Complex Man-made Worlds

In this paper, we describe the components of a robust algorithm for the detection of vanishing points in man-made environments. We designed our approach to work under quite general conditions (e.g., uncalibrated camera); and in contrast to several other approaches, the assumption of a dominant line-alignment w.r.t. the orthogonal axes of the world coordinate frame (Manhattan world) is not explicitly exploited. Our only premise is, that if a significant number of the imaged line segments meet very accurately in a point, this point is very likely to be a good candidate for a real vanishing point. For finding such points under a wide range of conditions, we propose a flexible algorithmic pipeline that combines accurate line detection techniques with robust statistical candidate initialization and refinement stages. The method was evaluated on a set of images exhibiting largely varying characteristics concerning image quality and scene complexity. Experiments show that the method, despite the variations, works in a stable manner and that its performance compares favorably with the state of the art.

Illumination insensitive recognition using eigenspaces

Variations in illumination can have a dramatic effect on the appearance of an object in an image. In this paper, we propose how to deal with illumination variations in eigenspace methods. We demonstrate that the eigenimages obtained by a training set under a single illumination condition (ambient light) can be used for recognition of objects taken under different illumination conditions. The major idea is to incorporate a gradient based filter bank into the eigenspace recognition framework. We show that the eigenimage coefficients are invariant to linear filtering (input and eigenimages are filtered with same filters). To achieve further illumination insensitivity we devised a robust procedure for coefficient recovery. The proposed approach has been extensively evaluated on a set of 4932 images and the results were compared to other approaches.

Mobile robot localization under varying illumination

Methods for mobile robot localization that use eigenspaces of panoramic snapshots of the environment are in general sensitive to changes in the illumination of the environment. Therefore, we propose an approach which achieves a reliable localization under severe illumination conditions. The method uses gradient filtering of the eigenspace. After testing the approach on images obtained by a mobile robot, we show that it outperforms the standard eigenspace-based recognition method.

Towards detection of orthogonal planes in monocular images of indoor environments

In this paper, we describe the components of a novel algorithm for the extraction of dominant orthogonal planar structures from monocular images taken in indoor environments. The basic building block of our approach is the use of vanishing points and vanishing lines imposed by the frequently observed dominance of three mutually orthogonal vanishing directions in man-made world. Vanishing points are found by an improved approach, taking no assumptions on known internal or external camera parameters. The problem of detecting planar patches is attacked using a probabilistic framework, searching for the maximum a posteriori probability (MAP) in a Markov Random Field (MRF). For this, we propose a novel formulation fusing geometric information obtained from vanishing points and features, such as rectangles and partial rectangles, together with a color-homogeneity criteria imposed by an image over-segmentation. The method was evaluated on a set of images exhibiting largely varying characteristics concerning image quality and scene complexity. Experiments show that the method, despite the variations, works in a stable manner and that its performance compares favorably to the state-of-the-art.

Efficient texture representation using multi-scale regions

This paper introduces an efficient way of representing textures using connected regions which are formed by coherent multi-scale over-segmentations. We show that the recently introduced covariance-based similarity measure, initially applied on rectangular windows, can be used with our newly devised, irregular structure-coherent patches; increasing the discriminative power and consistency of the texture representation. Furthermore, by treating texture in multiple scales, we allow for an implicit encoding of the spatial and statistical texture properties which are persistent across scale. The meaningfulness and efficiency of the covariance based texture representation is verified utilizing a simple binary segmentation method based on min-cut. Our experiments show that the proposed method, despite the low dimensional representation in use, is able to effectively discriminate textures and that its performance compares favorably with the state of the art.

Descriptor free visual indoor localization with line segments

We present a novel view on the indoor visual localization problem, where we avoid the use of interest points and associated descriptors, which are the basic building blocks of most standard methods. Instead, localization is cast as an alignment problem of the edges of the query image to a 3D model consisting of line segments. The proposed strategy is effective in low-textured indoor environments and in very wide baseline setups as it overcomes the dependency of image descriptors on textures, as well as their limited invariance to view point changes. The basic features of our method, which are prevalent indoors, are line segments. As we will show, they allow for defining an efficient Chamfer distance-based aligning cost, computed through integral contour images, incorporated into a first-best-search strategy. Experiments confirm the effectiveness of the method in terms of both, accuracy and computational complexity.

Weakly Supervised Group-Wise Model Learning Based on Discrete Optimization

In this paper we propose a method for the weakly supervised learning of sparse appearance models from medical image data based on Markov random fields (MRF). The models are learnt from a single annotated example and additional training samples without annotations. The approach formulates the model learning as solving a set of MRFs. Both the model training and the resulting model are able to cope with complex and repetitive structures. The weakly supervised model learning yields sparse MRF appearance models that perform equally well as those trained with manual annotations, thereby eliminating the need for tedious manual training supervision. Evaluation results are reported for hand radiographs and cardiac MRI slices.