Catálogo de publicaciones - libros

A traditional solution of area-based stereo uses some kind of windowed pixel intensity correlation. This approach suffers from discretization artifacts which corrupt the correlation value. We introduce a new correlation statistic, which is completely invariant to image sampling, moreover it naturally provides a position of the correlation maximum between pixels. Hereby we can obtain sub-pixel disparity directly from sampling invariant and highly discriminable measurements without any postprocessing of the discrete disparity map. The key idea behind is to represent the image point neighbourhood in a different way, as a response to a bank of Gabor filters. The images are convolved with the filter bank and the complex correlation statistic (CCS) is evaluated from the responses without iterations.

Vision is useful for the autonomous navigation of vehicles. In this paper the case of a vehicle equipped with multiple cameras with views is considered. The geometry and algebra of such a moving platform of cameras are considered. In particular we formulate and solve structure and motion problems for a few novel cases. There are two interesting minimal cases; three points in two platform positions and two points in three platform positions. We also investigate initial solutions for the case when image lines are used as features. In the paper is also discussed how classical algorithms such as intersection, resection and bundle adjustment can be extended to this new situation. The theory has been tested on synthetic and real data with promising results.

To analyze the performance of a vision-based tracking algorithm a good reference information is needed. Magnetic trackers are often used for this purpose, but the inevitable transformation of coordinate systems can result in notable alignment errors. This paper presents an approach for estimating the accuracy of various transformation models as well as individual model parameters. Performance is evaluated numerically and then tested on real data. Results show that the method can be successfully used to analyze the tracking error of free-moving objects.

We propose reliable outdoor object detection on mobile phone imagery from off-the-shelf devices. With the goal to provide both robust object detection and reduction of computational complexity for situated interpretation of urban imagery, we propose to apply the ’Informative Descriptor Approach’ on SIFT features (i-SIFT descriptors). We learn an attentive matching of i-SIFT keypoints, resulting in a significant improvement of state-of-the-art SIFT descriptor based keypoint matching. In the off-line learning stage, firstly, standard SIFT responses are evaluated using an information theoretic quality criterion with respect to object semantics, rejecting features with insufficient conditional entropy measure, producing both sparse and discriminative object representations. Secondly, we learn a decision tree from the training data set that maps SIFT descriptors to entropy values. The key advantages of informative SIFT (i-SIFT) to standard SIFT encoding are argued from observations on performance complexity, and demonstrated in a typical outdoor mobile vision experiment on the MPG-20 reference database.

This work proposes to learn visual encodings of attention patterns that enables sequential attention for object detection in real world environments. The system embeds a saccadic decision procedure in a cascaded process where visual evidence is probed at informative image locations. It is based on the extraction of information theoretic saliency by determining informative local image descriptors that provide selected foci of interest. The local information in terms of code book vector responses and the geometric information in the shift of attention contribute to recognition states of a Markov decision process. A Q-learner performs then performs search on useful actions towards salient locations, developing a strategy of action sequences directed in state space towards the optimization of information maximization. The method is evaluated in outdoor object recognition and demonstrates efficient performance.

We propose a method for finding stellate lesions in digitized mammograms based on the use of both local phase and local orientation information extracted from quadrature filter outputs. The local phase information allows efficient and fast separation between edges and lines and the local orientation estimates are used to find areas circumscribed by edges and with radiating lines. The method is incorporated in a computer-aided detection system and evaluation by FROC-curve analysis on a data set of 90 mammograms (45 pairs) yields a false positive rate of 0.3 per image at 90% sensitivity.

Since the publication of the comparative study done by Barron et al. on optical flow estimation, a race was started to achieve more and more accurate and dense velocity fields. For comparison a few synthetic image sequences has been used. The most complex of these is the Yosemite Flying sequence that contains both a diverging field, occlusion and multiple motions at the horizon. About 10 years ago it was suggested to remove the sky region because the correct flow used in earlier work was not found to be the real ground truth for this region. In this paper we present a study of the sky region in this test sequence, and discuss its usefulness for evaluation of optical flow estimation.

In this paper, we introduce a novel model-based visibility measure for geometric primitives called visibility map. It is simple to calculate, memory efficient, accurate for viewpoints outside the convex hull of the object and versatile in terms of possible applications. Several useful properties of visibility maps that show their superiority to existing visibility measures are derived. Various example applications from the automotive industry where the presented measure is used successfully conclude the paper.

Machine vision is today a well-established technology in industry where especially conveyer belt applications are successful. A related application area is the situation where a number of objects are located in a bin and each has to be picked from the bin. This problem is known as the automatic bin-picking problem and has a huge market potential due to the countless situations where bin-picking is done manually. In this paper we address a general bin-picking problem present at a large pump manufacturer, Grundfos, where many objects with circular openings are handled each day. We pose estimate the objects by finding the 3D opening based on the elliptic projections into two cameras. The ellipses from the two cameras are handled in a unifying manner using graph theory together with an approach that links a pose and an ellipse via the equation for a general cone. Tests show that the presented algorithm can estimate the poses for a large variety of orientations and handle both noise and occlusions.

In this paper we present a novel image-based 3D surface reconstruction technique that incorporates both reflectance and polarisation features into a variational framework. Our technique is especially suited for the difficult task of 3D reconstruction of rough metallic surfaces. An error functional consisting of several error terms related to the measured reflectance and polarisation properties is minimised in order to obtain a 3D reconstruction of the surface. We show that the combined approach strongly increases the accuracy of the surface reconstruction result, compared to techniques based on either reflectance or polarisation alone. We evaluate the algorithm based on synthetic ground truth data. Furthermore, we report 3D reconstruction results for a raw forged iron surface, thus showing the applicability of our method in real-world scenarios, here in the domain of quality inspection in the automotive industry.