Fig. 1. ECT capacitance sensor. (a) Structure of 12-electrode ECT sensor; (b) excitation state of electrode 1
Fig. 2. Three-dimensional distribution model of sensitivity between 1-7 electrode couple
Fig. 3. Imaging region convolution model
Fig. 4. Comparison of sensitivity gradients between 1-7 electrode couple before and after updating. (a) Before updating; (b) after updating
Fig. 5. Updating process of finite element sensitivity. (a) Updating process of finite element sensitivity S(i,j)(u) in sensitive field; (b) updating process of adjacent finite element sensitivity S(i,j)(u)
Fig. 6. Change curves of sensitivity gradient of 1-7 electrode couple under different iteration times. (a) Original sensitivity curves; (b) Once; (c) Two times; (d) three times; (e) four times; (f) five times
Fig. 7. Diagram of segmentation network and pixel distribution
Fig. 8. Sensitive field intensity changes under different iteration times and imaging comparison with Landweber algorithm
Fig. 9. IME values with different iteration times
Fig. 10. CORR values with different iteration times
Fig. 11. Images made by different algorithms
Fig. 12. IME histograms of different algorithms
Fig. 13. CORR histograms of different algorithms
Fig. 14. 12 electrode ECT system
Fig. 15. Experimental results of static imaging
Algorithm | Core | Floor | Fan | Four bubbles | Five bubbles | Six bubbles |
---|
Tikhonov | 0.2673 | 0.2638 | 0.4261 | 0.5640 | 0.6468 | 0.5830 | Landweber | 0.2472 | 0.2514 | 0.4151 | 0.4767 | 0.6200 | 0.5348 | Approximate L0 | 0.2530 | 0.2008 | 0.3650 | 0.3831 | 0.5837 | 0.6348 | Improved algorithm | 0.2436 | 0.1807 | 0.2874 | 0.3713 | 0.5154 | 0.5546 |
|
Table 1. IME results for different algorithms
Algorithm | Core | Floor | Fan | Four bubbles | Five bubbles | Six bubbles |
---|
Tikhonov | 0.8341 | 0.8766 | 0.8156 | 0.5625 | 0.5735 | 0.4697 | Landweber | 0.8166 | 0.8705 | 0.8262 | 0.5505 | 0.5791 | 0.5177 | Approximate L0 | 0.9125 | 0.8548 | 0.8290 | 0.7992 | 0.6462 | 0.5946 | Improved algorithm | 0.9152 | 0.8611 | 0.8462 | 0.8191 | 0.6955 | 0.6735 |
|
Table 2. CORR results for different algorithms