Performance Evaluation of 3D Local Feature Descriptors

Guo, Yulan; Bennamoun, Mohammed; Sohel, Ferdous; Lü, Min; Wan, Jianwei; Zhang, Jun

doi:10.1007/978-3-319-16808-1_13

Cited by 13 publications

(24 citation statements)

References 43 publications

(90 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…FPFH [44], SHOT [45], USC [48] and Spin Images [30] all use different ideas to capture these properties. Unfortunately, the challenges of real data, such as the presence of noise, missing structures, occlusions or clutter significantly harm such descriptors [26]. Recent trends in data driven approaches have encouraged the researchers to harness deep learning to surmount these nuisances.…”

Section: Related Workmentioning

confidence: 99%

“…Final estimation of the rigid pose between fragment pairs can then be made efficiently by operating on the pool of pose predictions. tion performance [26], directly solving the final problem at hand is certainly more critical. Unfortunately, contrary to 2D descriptors, the current deeply learned 3D descriptors [56,20,19] are still not tailored for the task we consider, i.e.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

3D Local Features for Direct Pairwise Registration

Deng¹,

Birdal

Ilić³

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

115

View full text Add to dashboard Cite

We present a novel, data driven approach for solving the problem of registration of two point cloud scans. Our approach is direct in the sense that a single pair of corresponding local patches already provides the necessary transformation cue for the global registration. To achieve that, we first endow the state of the art PPF-FoldNet [19] auto-encoder (AE) with a pose-variant sibling, where the discrepancy between the two leads to pose-specific descriptors. Based upon this, we introduce RelativeNet, a relative pose estimation network to assign correspondence-specific orientations to the keypoints, eliminating any local reference frame computations. Finally, we devise a simple yet effective hypothesize-and-verify algorithm to quickly use the predictions and align two point sets. Our extensive quantitative and qualitative experiments suggests that our approach outperforms the state of the art in challenging real datasets of pairwise registration and that augmenting the keypoints with local pose information leads to better generalization and a dramatic speed-up.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

3D Local Features for Direct Pairwise Registration

Deng¹,

Birdal

Ilić³

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

115

View full text Add to dashboard Cite

show abstract

“…where (x, y) are the pixel coordinates in I and w (u, v) is the window patch at position (u, v (3) where Iu, Iv represent the spatial gradients of the image.…”

Section: D Detectorsmentioning

confidence: 99%

“…Given the importance of these methods for 3D data registration and classification applications, it is necessary to evaluate keypoint detectors and feature descriptors [1]. Most such evaluations have been presented in the context of reports comparing current methods to newly proposed techniques, although some studies dedicated to the evaluation of 3D keypoint detectors or feature descriptors have also been published [2][3][4]. However, such evaluations have been limited to a single domain, with 3D methods applied directly to 3D data [2][3][4][5][6], or 2D methods applied to multiple 2D projections of 3D data [7][8][9][10][11].…”

Section: Introductionmentioning

confidence: 99%

Performance evaluation of single and cross‐dimensional feature detection and description

2020

View full text Add to dashboard Cite

Three-dimensional local feature detection and description techniques are widely used for object registration and recognition applications. Although several evaluations of 3D local feature detection and description methods have already been published, these are constrained in a single dimensional scheme, i.e. either 3D or 2D methods that are applied onto multiple projections of the 3D data. However, cross-dimensional (mixed 2D and 3D) feature detection and description has yet to be investigated. Here, we evaluated the performance of both single and cross-dimensional feature detection and description methods on several 3D datasets and demonstrated the superiority of cross-dimensional over single-dimensional schemes. 2D / 3D Keypoint Detection and Feature Description Methods Keypoint DetectorsKeypoint detectors analyse the structure around a vertex or a pixel depending on the data domain (3D or 2D,

show abstract

“…The SHOT method, based on a signature of histograms of orientations [41], is one of the best methods for the extraction and matching of features from overlapping partial shapes [4,17]. Even so, it usually unavoidably introduces mismatches amongst the established putative point matches (PPMs).…”

Section: Introductionmentioning

confidence: 99%

Accurately estimating rigid transformations in registration using a boosting-inspired mechanism

Liu

Martin

et al. 2016

Pattern Recognition

View full text Add to dashboard Cite

Feature extraction and matching provide the basis of many methods for object registration, modeling, retrieval, and recognition. However, this approach typically introduces false matches, due to lack of features, noise, occlusion, and cluttered backgrounds. In registration, these false matches lead to inaccurate estimation of the underlying transformation that brings the overlapping shapes into best possible alignment. In this paper, we propose a novel boosting-inspired method to tackle this challenging task. It includes three key steps: (i) underlying transformation estimation in * Corresponding author. Email: yitian.zhao@bit.edu.cn 1 the weighted least squares sense, (ii) boosting parameter estimation and regularization via Tsallis entropy, and (iii) weight re-estimation and regularization via Shannon entropy and update with a maximum fusion rule. The process is iterated. The final optimal underlying transformation is estimated as a weighted average of the transformations estimated from the latest iterations, with weights given by the boosting parameters. A comparative study based on real shape data shows that the proposed method outperforms four other state-of-the-art methods for evaluating the established point matches, enabling more accurate and stable estimation of the underlying transformation.

show abstract

Performance Evaluation of 3D Local Feature Descriptors

Cited by 13 publications

References 43 publications

3D Local Features for Direct Pairwise Registration

3D Local Features for Direct Pairwise Registration

Performance evaluation of single and cross‐dimensional feature detection and description

Accurately estimating rigid transformations in registration using a boosting-inspired mechanism

Contact Info

Product

Resources

About