Author Affiliations
School of Electrical Engineering and Automation, Anhui University, Hefei, Anhui 230601, Chinashow less
Fig. 1. Linear motor mover position measurement system
Fig. 2. Fence stripe images of different kinds. (a) Image of periodic fence; (b) aperiodic fence image of equal stripe width; (c) aperiodic fence image of black-white; (d) aperiodic fence image of gray scale
Fig. 3. Measurement errors of different algorithms. (a) Measurement errors of five algorithms; (b) comparison of measurement errors of two algorithms
Fig. 4. Displacement detection results under different intensity noises
Fig. 5. Runtime of different algorithms
Fig. 6. Regression structure diagram of DNN
Fig. 7. Modeling process of DNN
Fig. 8. Accuracy check of models. (a) DNN model; (b) SVM model
Fig. 9. Number of iterations of GA
Fig. 10. Experiment platform of position detection
Fig. 11. Signals acquired by line-scanning camera
Fig. 12. Error curves under different illuminations
Fig. 13. Measurement results of different target images
Fig. 14. Tracking accuracy of different speeds
Parameter | Name | Minimumvalue | Maximumvalue |
---|
g | average graygradient | 3 | 11 | s | width standarddeviation | 3 | 9 |
|
Table 1. Design parameters of the fence image
Serialnumber | Averagegray gradient | Width standarddeviation | Meanerror /pixel |
---|
No.1 | 8.33 | 6.99 | 0.0286 | No.2 | 7.17 | 7.08 | 0.0054 | No.3 | 4.33 | 6.49 | 0.1032 | No.4 | 8.25 | 6.29 | 0.0102 | No.5 | 6.04 | 7.39 | 0.0259 | No.6 | 8.31 | 7.56 | 0.0204 | … | … | … | … | No.659 | 6.45 | 7.55 | 0.0513 | No.660 | 5.34 | 7.04 | 0.0783 |
|
Table 2. Output of different fence images
Parameter | Name | Optimized result |
---|
g | average gray gradient | 10.98 | s | width standard deviation | 4.96 |
|
Table 3. Optimized design parameters