Author Affiliations
1College of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China2Fujian Science & Technology Innovation Laboratory for Optoelectronic Information of China, Fuzhou, Fujian 350116, China3School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362200, Chinashow less
Fig. 1. Segmentation result of the image (a) Original image; (b) processing result of the Otsu method; (c) gray histogram and threshold
Fig. 2. Acquisition of the peak information of image. (a) Gray histogram and its gradient; (b) cumulant of the gray histogram; (c) gradient cumulant of the gray histogram
Fig. 3. Normalized k(t) change curve with the gray-scale histogram
Fig. 4. Segmentation result of the electrowetting image 1. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshold
Fig. 5. Segmentation result of the electrowetting image 2. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) gray histogram and threshold
Fig. 6. Segmentation result of the electrowetting image 3. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshold
Fig. 7. Segmentation result of the fabric defect image 1. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshold
Fig. 8. Segmentation result of the fabric defect image 2. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshol
Fig. 9. Segmentation result of the welding defect image. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshold
Fig. 10. Segmentation result of the wood defect image. (a) Original image; (b) VE method; (c) Otsu method; (d) WOV method; (e) EW method; (f) GW method; (g) grayscale histogram and threshold
Application | Otsu | VE | WOV | EW | GW |
---|
Image 4 | 0.0353 | 0.0304 | 0.9182 | 0.8451 | 0.0359 | Image 5 | 0.8168 | 0.8168 | 0.8258 | 0.8148 | 0.0123 | Image 6 | 0.8751 | 0.8741 | 0.8787 | 0.8711 | 0.0076 | Image 7 | 0.5687 | 0.0008 | 0.0015 | 0.9984 | 0.0006 | Image 8 | 0.5091 | 0.0048 | 0.0048 | 0.9952 | 0.0014 | Image 9 | 0.3441 | 0.9873 | 0.9974 | 0.9974 | 0.0011 | Image 10 | 0.2567 | 0.0008 | 0.0026 | 0.0006 | 0.0002 | Average value | 0.4865 | 0.3879 | 0.5184 | 0.7889 | 0.0084 |
|
Table 1. Misclassification values of 5 methods
Application | Otsu | VE | WOV | EW | GW |
---|
Image 4 | 0.6907 | 0.7218 | 0.0033 | 0.0826 | 0.5436 | Image 5 | 0.0583 | 0.0583 | 0.0577 | 0.0584 | 0.7562 | Image 6 | 0.0207 | 0.0208 | 0.0207 | 0.0208 | 0.7093 | Image 7 | 0.0027 | 0.6588 | 0.0357 | 0.0000 | 0.7089 | Image 8 | 0.0093 | 0.0016 | 0.0016 | 0.0000 | 0.7066 | Image 9 | 0.0076 | 0.0026 | 0.0000 | 0.0000 | 0.5903 | Image 10 | 0.0225 | 0.8820 | 0.5669 | 0.9126 | 0.9653 | Average value | 0.1160 | 0.3351 | 0.0980 | 0.1535 | 0.7115 |
|
Table 2. Defect segmentation rates of 5 methods