Upright orientation of 3D shapes via tensor rank minimization

Wang, Weiming; Liu, Xiuping; Liu, Ligang

doi:10.1007/s12206-014-0604-6

Cited by 11 publications

(14 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(a) (b) Static Stability The upright orientation, which can be obtained by the state-of-art-methods [8,35], of the model defines its supporting plane. The points of the model on the supporting plane are named as supporting points.…”

Section: Structural Strengthmentioning

confidence: 99%

Cross section-based hollowing and structural enhancement

Wang

Qian

et al. 2017

Vis Comput

Self Cite

View full text Add to dashboard Cite

Recently, 3D printing has become a powerful tool for personal fabrication. However, the price of some materials are still high which limits its applications in home users. To optimize the volume of the model, while not largely affecting the strength of the objects, researchers propose algorithms to divide the model with different kind of light weight structures, such as frame structure, honeycomb-cell structure, truss structure, medial axis tree, etc. However, these algorithms are not suitable for the model whose internal space needs to be reused. In addition, the structural strength and static stability of the models, obtained with modern 3D model acquirement methods, are not guaranteed. In consequence, some models are too fragile to print, and cannot be survived in daily usage, handling, and transportation, or cannot stand in a stable.To handle the mentioned problems, an algorithm system is proposed based on cross sections in this work. The structural weak cross sections are enhanced and structural strong cross sections are adaptively hollowed to meet a given structural strength, static stability, printability, etc., while the material usage is minimized. The proposed algorithm system has been tested on several typical 3D models. The experimental results demonstrate the effectiveness and practicability of our system.

show abstract

Section: Structural Strengthmentioning

confidence: 99%

Cross section-based hollowing and structural enhancement

Wang

Qian

et al. 2017

Vis Comput

Self Cite

View full text Add to dashboard Cite

show abstract

“…In Wang et al . [31], a method is proposed by minimizing the tensor rank of the 3D shape's voxel representation. Both methods can handle shapes that have some kinds of symmetries.…”

Section: Related Workmentioning

confidence: 99%

Upright orientation of 3D shapes with Convolutional Networks

Liu¹,

Zhang²,

Liu³

2016

Graphical Models

Self Cite

View full text Add to dashboard Cite

Posing objects in their upright orientations is the very first step of 3D shape analysis. However, 3D models in existing repositories may be far from their right orientations due to various reasons. In this paper, we present a data-driven method for 3D object upright orientation estimation using 3D Convolutional Networks (Con-vNets), and the method is designed in the style of divide-and-conquer due to the interference effect. Thanks to the public big 3D datasets and the feature learning ability of ConvNets, our method can handle not only man-made objects but also natural ones. Besides, without any regularity assumptions, our method can deal with asymmetric and several other failure cases of existing approaches. Furthermore, a distance based clustering technique is proposed to reduce the memory cost and a test-time augmentation procedure is used to improve the accuracy. Its efficiency and effectiveness are demonstrated in the experimental results.

show abstract

“…Low Rank [69] intuitive, relatively robust Tensor Rank [70] capturing global symmetry, relatively robust Other Applications CMM [71] local controllability LBC [72] local controllability Skeleton Extraction [73] robust to noise and outlier 3D Printing [74] reduce the material largely LRSCPK [75] sharp feature preserving Point Cloud Compression [76] high compression ratio, robust to noise sharp feature preserving, robust to noise and outlier, and Reconstruction [16] unifying geometry and connectivity (2). Instead, Zhang et al [15] adopt the sparsity of face normals differences and propose a two-phase method including f ace normal filtering and vertex updating.…”

Section: Mesh Denoisingmentioning

confidence: 99%

“…It is very natural to generalize this method in 3D space to construct three-order tensor (multidimensional array) with volume of the 3D model, i.e., the three-order tensor ought to have a low rank behavior. Wang et al [70] construct this three-order tensor using the bounding box of the 3D model since the bounding box parallels the coordinate planes and contains the whole model. By translating the barycenter of the input model to the origin of the coordinate system, they just need to find and optimal rotation matrix R to align the model with three axes by following optimization model…”

Section: Upright Orientationmentioning

confidence: 99%