Watertight Multi-view Reconstruction Based on Volumetric Graph-Cuts

Sormann, Mario; Zach, Christopher; Bauer, Joachim; Karner, Konrad; Bishof, Horst

doi:10.1007/978-3-540-73040-8_40

Cited by 16 publications

(8 citation statements)

References 21 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, there is no easy handling of occlusions and visibility problems. Even if building a depth-map is faster nowadays, the fusion part implies a heavy computational cost [7], which hinders the use of such methods for our application.…”

Section: A Surface-based Methodsmentioning

confidence: 99%

Progressive 3D reconstruction of unknown objects using one eye-in-hand camera

Walck¹,

Drouin²

2009

2009 IEEE International Conference on Robotics and Biomimetics (ROBIO)

View full text Add to dashboard Cite

Abstract-This paper presents a complete 3D-reconstruction method optimized for online object modeling in the context of object grasping by a robot hand. The proposed solution is based on images captured by an eye-in-hand camera mounted on the robot arm and is an original combination of classical but simplified reconstruction methods. The different techniques used form a process that offers fast, progressive and reactive reconstruction of the object.

show abstract

Section: A Surface-based Methodsmentioning

confidence: 99%

Progressive 3D reconstruction of unknown objects using one eye-in-hand camera

Walck¹,

Drouin²

2009

2009 IEEE International Conference on Robotics and Biomimetics (ROBIO)

View full text Add to dashboard Cite

show abstract

“…Most of methods need some prior information as auxiliary to get more precise results. According to the taxonomy of Seitz et al [4], MVS algorithms can be divided into four categories: 3D volumetric approaches [5,6], surface evolution techniques [7,8], algorithms that compute and merge depth maps [9,10], and techniques that grow regions or surfaces starting from a set of extracted features or seed points [3,11], our algorithm apparently falls into the last category.…”

Section: Related Workmentioning

confidence: 99%

Scaled-neighborhood Patches Fusion for Multi-view Stereopsis

Dong

et al. 2016

MATEC Web of Conferences

View full text Add to dashboard Cite

Abstract. In this paper, we present a multi-view stereo reconstruction approach which fuses scaled-neighborhood information. PMVS proposed by Furukawa is one of the most excellent algorithms, and it has a good performance on many datasets both the accuracy and the completeness. However, there are still further improvements on this algorithm. PMVS cannot perform well in the presence of slanted surfaces, which are usually imaged at oblique angles. According to these aspects, on the one hand we propose to estimate the initial normal of every seed patch via fitting quadrics with scaled-neighborhood patches, which greatly improves the accuracy of the normal. On the other hand, we present to compute scaled-window for the further optimization based on texture. And it has been tested that employing the scaled-window will dramatically smooth the surfaces and enhance the reconstruction precision.

show abstract

“…Depending on size, number and shape of sampling geometry, our approach simulates exposing the scanned surface to a fluid pushing through the model. The idea of generating watertight models from previously generated datasets has been documented before [17], [16], so the novelties here are:…”

Section: Algorithmmentioning

confidence: 99%

Towards autonomous airborne mapping of urban environments

Adler¹,

Xiao²

2012

2012 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)

View full text Add to dashboard Cite

Abstract-This work documents our progress on building an unmanned aerial vehicle capable of autonomously mapping urban environments. This includes localization and tracking of the vehicle's pose, fusion of sensor-data from onboard GNSS receivers, IMUs, laserscanners and cameras as well as realtime path-planning and collision-avoidance. Currently, we focus on a physics-based approach to computing waypoints, which are subsequently used to steer the platform in three-dimensional space. Generation of efficient sensor trajectories for maximized information gain operates directly on unorganized point clouds, creating a perfect fit for environment mapping with commonly used LIDAR sensors and time-of-flight cameras. We present the algorithm's application to real sensor-data and analyze its performance in a virtual outdoor scenario.

show abstract

Watertight Multi-view Reconstruction Based on Volumetric Graph-Cuts

Cited by 16 publications

References 21 publications

Progressive 3D reconstruction of unknown objects using one eye-in-hand camera

Progressive 3D reconstruction of unknown objects using one eye-in-hand camera

Scaled-neighborhood Patches Fusion for Multi-view Stereopsis

Towards autonomous airborne mapping of urban environments

Contact Info

Product

Resources

About