Active Visual Object Reconstruction using D-, E-, and T-Optimal Next Best Views

Wenhardt, Stefan; Deutsch, Benjamin; Angelopoulou, Elli; Niemann, Heinrich

doi:10.1109/cvpr.2007.383363

Cited by 46 publications

(36 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The previous pipeline has become a standard for Multi-view Stereo Reconstruction, and it was implemented in several versions, both structured as webservices or closed systems, where the user must provide all the images which were acquired, and the interaction is limited to the possibility to tune the parameters of the various steps of the reconstruction. Some attempts have been done to break this paradigm, by proposing methods to assist the user during the acquisition, essentially by estimating a set of missing view to be suggested to the user [16], [17], [18]. Our approach differentiates from the latter because there is no assumption on the object shape and the acquisition setup, and the object of interest and the suggested view(s) are generated only by analyzing the sparse reconstruction of the scene.…”

Section: Related Workmentioning

confidence: 99%

Assisted Multi-view Stereo Reconstruction

Dellepiane

Cavarretta

Cignoni

et al. 2013

2013 International Conference on 3D Vision

View full text Add to dashboard Cite

Abstract-Multi-view stereo reconstruction methods can provide impressive results in a number of applications. Nevertheless, when trying to apply the state-of-the-art methods in the case of a more structured 3D acquisition, the lack of feedback on the quality of the reconstruction during the photo shooting can be problematic. In this paper we present a framework for the assisted reconstruction from images of real objects. The framework is able to provide, in quasi-realtime, a sparse reconstruction of the scene, so that the user is able to spot the missing or problematic parts. Moreover, the framework is able to separate the object of interest from the background and suggests missing points of view to the user, without any previous knowledge of the shape of the scene and the acquisition path. This is obtained by analyzing the sparse reconstruction and the connection between the reconstructed points and the input images. The framework has been tested on a variety of practical cases, and it has proved to be effective not only to obtain more complete reconstructions, but also to reduce the number of images needed and the processing time for dense reconstruction.

show abstract

Section: Related Workmentioning

confidence: 99%

Assisted Multi-view Stereo Reconstruction

Dellepiane

Cavarretta

Cignoni

et al. 2013

2013 International Conference on 3D Vision

View full text Add to dashboard Cite

show abstract

“…As input data we use predefined image sequences as well as a planned sequence produced by the information-theoretic NBV planning approach described in [5]. Figure 2 shows the experimental setup.…”

Section: Experimental Comparison Of Klt Gklt and Gklt 3d Trackingmentioning

confidence: 99%

“…NBV planning uses these additional possibilities to achieve defined reconstruction goals. The NBV planning method in [5] uses an extended Kalman filter to compute 3D reconstructions of tracked features and determines the next best view by applying an information-theoretic quality criterion and visibility constraints. We track 496 features and use the short test sequence as the initial sequence of the planning procedure.…”

Section: Comparison Within An Information-theoretic Approach For Nextmentioning

confidence: 99%

See 1 more Smart Citation

Combined GKLT Feature Tracking and Reconstruction for Next Best View Planning

Trummer

Munkelt

Denzler

2009

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Guided Kanade-Lucas-Tomasi (GKLT) tracking is a suitable way to incorporate knowledge about camera parameters into the standard KLT tracking approach for feature tracking in rigid scenes. By this means, feature tracking can benefit from additional knowledge about camera parameters as given by a controlled environment within a next-best-view (NBV) planning approach for three-dimensional (3D) reconstruction. We extend the GKLT tracking procedure for controlled environments by establishing a method for combined 2D tracking and robust 3D reconstruction. Thus we explicitly use the knowledge about the current 3D estimation of the tracked point within the tracking process. We incorporate robust 3D estimation, initialization of lost features, and an efficient detection of tracking steps not fitting the 3D model. Our experimental evaluation on real data provides a comparison of our extended GKLT tracking method, the former GKLT, and standard KLT tracking. We perform 3D reconstruction from predefined image sequences as well as within an information-theoretic approach for NBV planning. The results show that the reconstruction error using our extended GKLT tracking method can be reduced up to 71% compared to standard KLT and up to 39% compared to the former GKLT tracker.

show abstract

“…However, scanning and registration in this case highly depends on having the human body shape as a prior. Some NBV techniques [Wenhardt et al 2007;Dunn and Frahm 2009] build 3D models from images, based on camera movements and on utilizing the model's covariance structure as well as texture appearance. In our work, we instead aim at high fidelity 3D scanning of unknown objects that allow automatic reconstruction of objects with rich geometric details and a complex topology.…”

Section: Related Workmentioning

confidence: 99%

Quality-driven poisson-guided autoscanning

Wu¹,

Sun²,

Long³

et al. 2014

ACM Trans. Graph.

View full text Add to dashboard Cite

Figure 1: Our robot-based, Poisson-guided autoscanner can progressively, adaptively, and fully automatically generate complete, high quality, and high fidelity scan models. AbstractWe present a quality-driven, Poisson-guided autonomous scanning method. Unlike previous scan planning techniques, we do not aim to minimize the number of scans needed to cover the object's surface, but rather to ensure the high quality scanning of the model. This goal is achieved by placing the scanner at strategically selected Next-Best-Views (NBVs) to ensure progressively capturing the geometric details of the object, until both completeness and high fidelity are reached. The technique is based on the analysis of a Poisson field and its geometric relation with an input scan. We generate a confidence map that reflects the quality/fidelity of the estimated Poisson iso-surface. The confidence map guides the generation of a viewing vector field, which is then used for computing a set of NBVs. We applied the algorithm on two different robotic platforms, a PR2 mobile robot and a one-arm industry robot. We demonstrated the advantages of our method through a number of autonomous high quality scannings of complex physical objects, as well as performance comparisons against state-of-the-art methods.

show abstract

Active Visual Object Reconstruction using D-, E-, and T-Optimal Next Best Views

Cited by 46 publications

References 7 publications

Assisted Multi-view Stereo Reconstruction

Assisted Multi-view Stereo Reconstruction

Combined GKLT Feature Tracking and Reconstruction for Next Best View Planning

Quality-driven poisson-guided autoscanning

Contact Info

Product

Resources

About