Author Affiliations
1School of Mechanical Engineering, Jiangnan University, Wuxi, Jiangsu 214122, China2Jiangsu Key Laboratory of Advanced Food Manufacturing Equipment & Technology, Wuxi, Jiangsu 214122, China;3School of Internet of Things Engineering, Jiangnan University, Wuxi, Jiangsu 214122, Chinashow less
Fig. 1. Comparison results before and after preprocessing. (a) Image pyramid; (b) original image; (c) image after preprocessing
Fig. 2. Grayscale variation image of local section of mesh. (a) Local image of mesh fabric; (b) distribution curve of gray value in local linear neighborhood
Fig. 3. Mesh fabric images with three different light levels. (a) Sample 1; (b) sample 2; (c) sample 3
Fig. 4. Results of segmenting sample 1 with different thresholds and its partial enlarged views. (a) T=20; (b) T=50; (c) T=100; (d) proposed algorithm; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Fig. 5. Results of segmenting sample 2 with different thresholds and its partial enlarged views. (a) T=20; (b) T=50; (c) T=100; (d) proposed algorithm; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Fig. 6. Results of segmenting sample 3 with different thresholds and its partial enlarged views. (a) T=20; (b) T=50; (c) T=100; (d) proposed algorithm; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Fig. 7. Results of segmenting sample 1 with different algorithms and its partial enlarged views. (a) Shading threshold algorithm; (b) marking watershed algorithm; proposed algorithm (c) before preprocessing, (d) after preprocessing; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Fig. 8. Results of segmenting sample 2 with different algorithms and its partial enlarged segmenting. (a) Shading threshold algorithm; (b) marking watershed algorithm; proposed algorithm (c) before preprocessing, (d) after preprocessing; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Fig. 9. Results of segmenting sample 3 with different algorithms and its partial enlarged views. (a) Shading threshold algorithm; (b) marking watershed algorithm; proposed algorithm (c) before preprocessing, (d) after preprocessing; (e) partial enlargement of Fig. (a); (f) partial enlargement of Fig. (b); (g) partial enlargement of Fig. (c); (h) partial enlargement of Fig. (d)
Algorithm | Sample 1 | Sample 2 | Sample 3 |
---|
Number ofsegmentationerrors | Segmentationerrorrate/% | Number ofsegmentationerrors | Segmentationerrorrate/% | Number ofsegmentationerrors | Segmentationerrorrate/% |
---|
Mark watershedalgorithm | 41 | 11.0 | 21 | 4.8 | 10 | 3.3 | Proposedalgorithms | Beforepreprocessing | 162 | 43.0 | 183 | 41.1 | 74 | 24.3 | Afterpreprocessing | 1 | 0.3 | 1 | 0.2 | 0 | 0 |
|
Table 1. Number of mesh segmentation errors and segmentation error rate
Algorithm | Mark watershedalgorithm | Proposed algorithm |
---|
Before preprocessing | After preprocessing |
---|
Number of meshes | | 34932 | | Number of mesh segmentation errors | 2061 | 11741 | 84 | Segmentation error rate/% | 5.90 | 33.61 | 0.24 |
|
Table 2. Comparison of mesh segmentation results of different algorithms