Laser & Optoelectronics Progress, Volume. 58, Issue 14, 1410007(2021)
Deep Convolution Neural Network Method for Skew Angle Detection in Text Images
Figures & Tables(11)
Fig. 3. The example 1 of text image quality skew angle multi-stage and multi classification detection
Fig. 4. The example 2 of text image quality skew angle multi-stage and multi classification detection
Fig. 5. The example 3 of text image quality skew angle multi-stage and multi classification detection
Table 1. Tilt angle class and angle range value
View tableTable 1. Tilt angle class and angle range value
Tilt classification Tilt angle range Tilt classification Tilt angle range L1 [0, 30°) L7 [0, -30°) L2 [30°, 60°) L8 [-30°, -60°) L3 [60°, 90°) L9 [-60°, -90°) L4 [90°, 120°) L10 [-90°, -120°) L5 [120°, 150°) L11 [-120°, -150°) L6 [150°, 180°) L12 [-150°, -180°)
Table 2. Network parameters
View tableTable 2. Network parameters
Layer name Filter size Stride Padding Conv1 3×3×1×64 2 1 Conv2 3×3×64×64 2 1 Max Pooling 2×2 2 0 Conv3 3×3×64×64 2 1 Max Pooling 2×2 2 0 Conv4 3×3×64×64 2 1 Max Pooling 2×2 2 0 Conv5 3×3×64×64 2 1 Conv6 3×3×64×10 2 1
Table 3. Detection results of text image tilt angle classification (12 classifications)
View tableTable 3. Detection results of text image tilt angle classification (12 classifications)
Performance index One stage structure Multistage structure ( Fig.2 )( Fig.3 )( Fig.4 )( Fig.5 )Accuracy 0.982 0.955 0.962 0.956 Recall 0.976 0.966 0.954 0.962 Precision 0.982 0.955 0.962 0.956 F1_Score 0.978 0.965 0.951 0.958 Train time /h 4.321 9.654 9.461 9.622 Test time /ms 7.139 8.112 8.121 8.110
Table 4. Detection results of text image tilt angle classification (24 classifications)
View tableTable 4. Detection results of text image tilt angle classification (24 classifications)
Performance index Multistage structure ( Fig.2 ’)( Fig.3 ’)( Fig.4 ’)Accuracy 0.945 0.943 0.949 Recall 0.946 0.949 0.952 Precision 0.934 0.936 0.937 F1_Score 0.940 0.942 0.944 Train time/h 16.785 16.768 16.642 Test time/ms 13.609 13.434 13.536
Table 5. Comparison of text image recognition before and after skew correction
View tableTable 5. Comparison of text image recognition before and after skew correction
Tilt angle classification Precision of ASTER Precision of MORAN Precision of CRNN Before tilt correction After tilt correction Before tilt correction After tilt correction Before tilt correction After tilt correction L1 0.966 0.974 0.977 L2 0.434 0.964 0.454 0.965 0.464 0.960 L3 0.171 0.961 0.161 0.966 0.181 0.956 L4 0.205 0.963 0.015 0.953 0.115 0.957 L5 0 0.959 0 0.959 0 0.959 L6 0 0.959 0 0.958 0 0.966 L7 0.807 0.964 0.817 0.947 0.837 0.959 L8 0.217 0.958 0.247 0.958 0.247 0.951 L9 0.015 0.946 0.044 0.951 0.055 0.966 L10 0 0.966 0 0.942 0 0.958 L11 0 0.962 0 0.945 0 0.957 L12 0 0.955 0 0.961 0 0.941 L13 0 0.966 0 0.974 0 0.977
Tools
Get Citation
Copy Citation Text
Congzhou Guo, Ke Li, Yikun Zhu, Xiaochong Tong, Xiwen Wang. Deep Convolution Neural Network Method for Skew Angle Detection in Text Images[J]. Laser & Optoelectronics Progress, 2021, 58(14): 1410007
Paper Information
Category: Image Processing
Received: Sep. 22, 2020
Accepted: Nov. 14, 2020
Published Online: Jun. 30, 2021
The Author Email: Congzhou Guo (czguo0618@sina.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20