Laser & Optoelectronics Progress, Volume. 56, Issue 6, 061006(2019)
Lightweight Staff Removal Method Based on Multidimensional Local Binary Pattern and XGBoost
Figures & Tables(8)
Fig. 2. Example of score used in experiment. (a) Input score; (b) ground-truth label
Fig. 5. Results of score part. (a) Input score; (b) ground-truth label; (c) processing result
Table 1. F-measure of different window sizes%
View tableTable 1. F-measure of different window sizes%
Size TS1 TS2 TS3 Whole 7×7 94.13 97.93 95.76 95.94 9×9 96.02 98.12 96.57 96.90 11×11 95.88 98.06 96.52 96.82 13×13 96.26 98.15 96.52 96.97 15×15 95.98 98.11 96.43 96.84 17×17 94.97 98.08 95.85 96.30 7×7+9×9 95.98 98.06 96.21 96.75 9×9+11×11 95.92 98.10 96.38 96.80 11×11+13×13 96.03 98.07 96.25 96.78 9×9+13×13 96.16 98.34 96.65 97.05 13×13+17×17 96.09 98.22 96.17 96.82
Table 2. F-measure comparison of different classifiers%
View tableTable 2. F-measure comparison of different classifiers%
Classifier TS1 TS2 TS3 Whole SVM 95.56 96.27 96.45 96.09 Adaboost 95.31 96.10 96.33 95.91 KNN 95.89 97.02 96.20 96.37 RF 96.11 97.54 96.36 96.67 XGBoost 96.43 98.36 96.79 97.19
Table 3. F-measure comparison of different methods%
View tableTable 3. F-measure comparison of different methods%
Method TS1 TS2 TS3 Whole TAU 85.72 81.72 82.29 83.01 NUS 69.85 96.25 67.43 75.24 NUASI-lin 94.99 94.86 94.00 94.29 NUASI-skel 94.25 93.80 92.92 93.34 LRDE 97.73 96.86 96.98 97.14 INESC 89.29 97.72 88.52 91.01 Proposed approach 96.43 98.36 96.79 97.19
Tools
Get Citation
Copy Citation Text
Tianlong Wu, Qiang Li, Xin Guan. Lightweight Staff Removal Method Based on Multidimensional Local Binary Pattern and XGBoost[J]. Laser & Optoelectronics Progress, 2019, 56(6): 061006
Paper Information
Category: Image Processing
Received: Sep. 26, 2018
Accepted: Oct. 25, 2018
Published Online: Jul. 30, 2019
The Author Email: Guan Xin (guanxin@tju.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20