Acta Optica Sinica, Volume. 38, Issue 9, 0915002(2018)
Occlusion Boundary Detection of Deep Image by Using Spectral Clustering
Figures & Tables(6)
Fig. 1. Overall process of occlusion boundary detection of depth image based on spectral clustering
Fig. 3. Detection results of occlusion boundary based on different features. (a) Depth image; (b) Ground Truth;(c) method in Ref.[9] combined with related features; (d) method in Ref. [9] combined with joint features;(e) proposed method combined with maximal depth difference feature;(f) proposed method combined with proposed feature
Fig. 4. Detection results of occlusion boundary based on different methods. (a) Depth image; (b) Ground Truth;(c) method in Ref. [7]; (d) method in Ref. [10]; (e) method in Ref. [11]; (f) proposed method
Table 1. Quantitative analysis of occlusion boundary detection results based on different features
View tableTable 1. Quantitative analysis of occlusion boundary detection results based on different features
Method Feature Depth image N gtN dN tR p /%R r /%F -scoreMethod in Ref. [9] Existed features Bunny 732 731 724 99.04 98.91 98.97 Rocker 656 641 640 99.84 97.56 98.69 Joint features Bunny 732 732 727 99.32 99.32 99.32 Rocker 656 643 643 99.84 97.87 98.85 Proposed method Maximal depth difference feature Bunny 732 757 691 91.28 94.40 92.81 Rocker 656 598 598 1.00 91.16 95.37 Proposed feature Bunny 732 771 701 90.92 95.77 93.28 Rocker 656 615 614 99.84 93.60 96.62
Table 2. Quantitative evaluation results of different occlusion boundary detection methods
View tableTable 2. Quantitative evaluation results of different occlusion boundary detection methods
Method R ap /%R ar /%F -scoreMethod in Ref. [7] 52.18 50.35 0.5125 Method in Ref. [10] 95.38 96.07 0.9572 Method in Ref. [11] 95.92 91.68 0.9379 Proposed method 93.60 93.33 0.9346
Tools
Get Citation
Copy Citation Text
Shihui Zhang, Meng Yang, Lijian Dong. Occlusion Boundary Detection of Deep Image by Using Spectral Clustering[J]. Acta Optica Sinica, 2018, 38(9): 0915002
Paper Information
Category: Machine Vision
Received: Dec. 29, 2017
Accepted: Apr. 16, 2018
Published Online: May. 9, 2019
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20