Model-based Recovery and Tracking

First Pose computation, model-based tracking and path planning for augmented reality

ABSTRACT

This work presents a study of the existing optical markerless tracking methods and provides some improvements for some of them. The proposed improvements are focused on industrial environments, which is a difficult challenge due to the lack of texture in these scenes. The response of model based monocular optical markerless tracking methods is jeopardized for untextured scenes, so this work proposes a 3D object recognition method that uses geometric properties instead of texture to initialize the tracking, as well as a markerless tracking method that uses multiple visual cues to update the tracking. Additionally, an augmented reality system has been developed to help in disassembly operations. This serves as a tool to validate the proposed methods and shows a real world applicability of them.

Concerning to the 3D recognition of untextured 3D models in a single image, the method is capable of recovering a 3D pose in a few hundred of milliseconds, which is a difficult challenge using this type of model. Because the target 3D models lack of texture, their geometry features have been used as a basis for their 3D recognition. An automatic process extracts the junctions and contours of the model, replacing the user interaction. Junctions will provide us an efficient mechanism to generate candidate matches, while contours will select the correct match based on a robust shape similarity evaluation. Our method only requires the mesh of the model as input, since the rest of the process is done automatically. We demonstrate the behavior of our approach against a variety of real scenes and models. Moreover, we explain how to face the first pose problem in a robust way using a history of votes. We also present a study of the method parameterisation, describing the influence of each parameter. This method can be used in both 3D object recognition and first pose estimation for markerless augmented reality applications.

PAPERS

Books:

• Álvarez, H., Study of Augmented Reality Methods for Real Time Recognition and Tracking of Untextured 3D Models in Monocular Images (ISBN 84-8081-154-4), PhD Thesis Dissertation in December 2011. Ed. Universidad de Navarra (January 2012). (EG Digital Library pdf).

Journals

• Álvarez, H., and Borro, D., “Junction assisted 3D pose retrieval of untextured 3D models in monocular images”, Computer Vision and Image Understanding, Vol. 117, No. 10, pp. 1204-1214. October 2013. (pdf).

 Conferences

• Álvarez, H., Aguinaga, I., and Borro, D., “Providing Guidance for Maintenance Operations Using Automatic Markerless Augmented Reality System”, Proceedings of the 10th IEEE International Symposium on Mixed and Augmented Reality (ISMAR 2011), pp. 181-190 (ISBN: 978-1-4577-2183-0). Basel, Switzerland. October 26-29, 2011. (pdf, video).

VIDEOS

• Video01: 3D object recognition (Youtube link)

• Video02: Markerless Augmented Reality for Maintenance (Youtube link)

• Video03: 3D object recognition based on Junctions (Youtube link)

RELATED PROJECTS