NUI Maynooth
Maynooth, Co. Kildare
guillaume.gales [at] nuim [dot] ie
PostDoc advisor John McDonald
Ph.D. in computer science from the University of Toulouse.
Ph.D. advisors: Sylvie CHAMBON and Alain CROUZIL.
This talk discusses a number of techniques for correspondence estimation between stereo image pairs, i.e. two images of the same scene taken from different positions. The problem is to identify pairs of pixels in the two images that are the projections of the same scene point. Although the human visual system performs this task with ease, developing algorithms for automatically computing correspondences is a challenging task. In particular, existing algorithms can fail in homogeneous areas, near depth discontinuities and occlusions or with a repetitive texture pattern. The first part of this talk focuses on seed propagation-based approaches that are a special case of local methods based computing an iterative solution, where the solution is initialised using a sparse set of reliable matches (the seeds). I introduce a reliability measure used by the propagation technique for finding the correct correspondent of a pixel, providing robustness in the context of the above difficulties. This measure takes into account an unambiguity term, a continuity term and a colour consistency term. It has the advantage of taking into account information from the other candidates, and leads, according to our experimental evaluation, to better results when compared to other methods based on a correlation score alone. In the second part of this talk I will present ongoing work in our group on stereo matching in urban environments. In particular we exploit the fact that images of such environments contain multiple planar elements. I will show how utilising this strong geometrical constraint allows us to automatically segment building facades in single images. Furthermore I show how this technique permits robust pixel matching in wide-baseline stereo pairs. Finally, I will discuss how we intend to apply this technique for the development of augmented reality applications.
Wednesday 21st March 2012, 3.00pm – 4.00pm, Computer Science Room (No. 1.42), Ground Floor, Callan Building, NUI Maynooth. If you would be interested in attending this seminar, please RSVP Joanna O’ Grady either via e-mail jogrady at eeng dot nuim dot ie.
This demo is based on Yanpeng Cao and John McDonald's paper: Improved feature extraction and matching in urban environments based on 3D viewpoint normalization. We provide here a demonstration of the 3D Viewpoint Normalization. Upload an image of a building, then hit process to see the result of the normalization. TBA.
Stereo matching is one of the main topics in computer vision.
It consists in finding in two images of a same scene, taken from different viewpoints, the pairs of pixels which are the projections of a same scene point.
Since the last twenty years, many local and global methods have been proposed to solve this problem.
More recently, according to a reference evaluation protocol in the community, region-based methods showed interesting result in small-baseline binocular stereo (where images are taken nearby).
The idea is to apply a colour segmentation algorithm on the images assuming that each pixel within a segment belongs to a same object surface. Then, the parameters of a surface model are computed, in the disparity space, for each segment according to initial disparities usually computed with a local method.
Finally, a global optimization is performed to refine the results.
A contribution of this thesis deals with a special kind of local method called seeds propagation.
The search area of a correspondent is reduced to the neighbourhoods of reliable matches called seeds.
This can help to reduce the computation time and to avoid some ambiguities.
However, the success of such a method depends on the choice of these seeds.
In this dissertation, we give a study of the seeds selection step.
We focus on feature points matching. These are special points in the image with interesting characteristics for a given application.
In our case, we need pixels that can be matched with high confidence.
We compare fourteen different well-known detectors linked to five correlation measures. Some of these measures are meant to be robust to one of the main challenge in stereo matching: depth discontinuities.
Besides, this study gives advice on the choice of the parameters of the different techniques to be able to find the best solutions according some given criteria. These parameters are estimated using machine learning.
Then, these seeds are used with two approaches of propagation and the results are evaluated.
Another contribution deals with a new region-based approach for dense stereo matching.
Different colour segmentations are used. Then, many instances of a surface model are computed for the different regions according to initial disparities selected randomly.
For each pixel, each instance gives a disparity value regarded as a vote.
Finally, the most voted value is selected as the final disparity.
This approach is relatively easy to implement and very effective giving competitive results among the state of the art.