Fig. 1. Sketch of the edge detection system
Fig. 2. Top view of probe and moving textile. (a) Normal position; (b) left shift; (c) right shift
Fig. 3. Flow chart of detecting edge
Fig. 4. Pixel distribution
Fig. 5. Original image
Fig. 6. Gray scale response curve of textile and background
Fig. 7. Grayscale response curve with only edge information
Fig. 8. Flow chart for solving edge starting point xk
Fig. 9. Flow chart for solving edge terminating point xj
Fig. 10. Experimental platform
Fig. 11. Creamy white carpet edge detection result. (a) Original grayscale; (b) after median filtering; (c) after algorithm post-processing
Fig. 13. Edge detection result of dark gray cotton cloth with hollowed-out and white stripes. (a) Original grayscale; (b) after median filtering; (c) after algorithm post-processing
Fig. 14. Edge detection result of dark gray cotton cloth with hollowed-out and white stripes in a light gray background. (a) Original grayscale; (b) after median filtering; (c) after algorithm post-processing
Fig. 15. Contrast among OTSU algorithm, lattice Boltzmann algorithm, and designed algorithm for edge detection. (a) Original grayscale; (b) after median filtering; (c) after OTSU algorithm post-processing; (d) after lattice Boltzmann algorithm post-processing; (e) after designed algorithm post-processing
Parameter | Description |
---|
Ambient temperature range | 10-25 ℃ | Ambient humidity (relative) | 30%-60% | Measuring height | 36 mm±2 mm | Experimental background | Uniform black light absorbent cloth with flannelette | Environmental light irradiation | Natural light in the laboratory | Light source type | Stripe white light source | Intensity of light source | Constant size, controlled by light source controller | Linear CCD model | TCD1209DG | CCD exposure time | 2 ms | Lens magnification | 0.9× | Algorithm computing platform | MATLAB 2013a | CPU | Intel Core2 Duo CPU | Operating system (OS) | Windows 7 |
|
Table 1. Experimental parameters of environment platform
Template | Noise-to-signalratio /dB | Processingtime /ms |
---|
5×1 | 38.4958 | 1.890 | 7×1 | 38.2115 | 2.160 | 9×1 | 37.6220 | 2.181 |
|
Table 2. Filtering data of 5×1, 7×1, and 9×1 templates
Fabric | Actual edge position(pixel ordinal number) | Detected edge position(pixel ordinal number) | Processing time /ms |
---|
Creamy white carpet | 1013 | 1016 | 0.264 | Dark gray carpet | 985 | 989 | 0.265 |
|
Table 3. High and low gray response fabric edge detection data
Fabric | Actual edge position | Detected edge position | Processing time /ms |
---|
Dark gray cotton cloth withhollow and white stripes | 1189 | 1186 | 0.279 |
|
Table 4. Edge detection data of dark gray cotton cloth with hollowed-out and white stripes
Background | Actual edge position | Detected edge position | Processing time /ms |
---|
Black background cloth | 1189 | 1186 | 0.279 | Light gray background cloth | 1190 | 1188 | 0.286 |
|
Table 5. Edge detection data for different background colors
Algorithm | Actual edge position | Detected edge position | Processing time /ms |
---|
OTSU | 1163 | 982 | 8.478 | Lattice Boltzmann | 1163 | 1155 | 10.658 | Designed | 1163 | 1165 | 0.266 |
|
Table 6. Edge detection data of different algorithms