|
|
|
Multi-Threshold Technique for Segmentation of Satellite Imagery for Feature Extraction
Segmentation Process
After the entire optimal threshold values are determined then input these values to the automatic segmentation produce program. This program is by using PCI Modeler to implement the automatic segmentation for entire image. The segment feature covers the entire image are urban, clear land, forest and water. All of these segments are fused together to produce a segmentation map as shown in figure 5.
Table 1. Minimum and maximum grey-level
values of each feature in different band
| Band (k) |
Min(i, k ) |
Max(i, k ) |
| Blue (1) |
Min(1, 1) = 73 |
Max(1,1) =124 |
|
Min(2, 1) = 80 |
Max(2, 1) = 149 |
|
Min(3, 1) = 56 |
Max(3, 1) = 70 |
|
Min(4, 1) = 59 |
Max(4, 1) = 68 |
| Green (2) |
Min(1, 2) = 30 |
Max(1, 2) = 58 |
|
Min(2, 2) = 43 |
Max(2, 2) = 96 |
|
Min(3, 2) = 19 |
Max(3, 2) = 30 |
|
Min(4, 2) = 19 |
Max(4, 2) = 25 |
| Red (3) |
Min(1, 3) = 24 |
Max(1, 3) = 64 |
|
Min(2, 3) = 58 |
Max(2, 3) = 160 |
|
Min(3, 3) = 5 |
Max(3, 3) = 19 |
|
Min(4, 3) = 5 |
Max(4, 3) = 14 |
| Near-Infrared (4) |
Min(1, 4) = 37 |
Max(1, 4) = 66 |
|
Min(2, 4) = 46 |
Max(2, 4) = 132 |
|
Min(3, 4) = 24 |
Max(3, 4) = 108 |
|
Min(4, 4) = 0 |
Max(4, 4) = 47 |
| Mid-Infrared (5) |
Min(1, 5) = 56 |
Max(1, 5) = 124 |
|
Min(2, 5) = 68 |
Max(2, 5) = 255 |
|
Min(3, 5) = 11 |
Max(3, 5) = 102 |
|
Min(4, 5 ) = 0 |
Max(4, 5 ) = 29 |
| Mid-Infrared (6) |
Min(1, 6) = 31 |
Max(1, 6) = 88 |
|
Min(2, 6) = 24 |
Max(2, 6) = 148 |
|
Min(3, 6) = 4 |
Max(3, 6) = 40 |
|
Min(4, 6 ) = 0 |
Max(4, 6 ) = 12 |
Table 2. Optimal threshold values of each feature in different band
Conclusion
In this paper, an automatic thresholding technique has been proposed for segmentation the multiband images and successfully tested with a Landsat TM image. The method is based on two important algorithms; pair features overlapping analysis and optimal threshold values selection. The results presented in this paper show the efficiency of the method for image segmentation.
References
- Chang, J. S., Liao, H. Y. M., Hor, M. K., Hsieh, J. W., and Chern, M. Y., New automatic multi-level threshold technique for segmentation of thermal images. Image and Vision Computing.,15:23-34(1997).
- Papamarkos, N., Strouthopoulos, C., and Andreadis, I., Multithresholding of color and grey-level images through a neural network technique. Image and Vision Computing.,18: 213-222 (2000).
- Sato, K., Nakajima, M., and Hoshi, T., Thresholding operation for extraction of mangrove forest with TM data of LANDSAT 5. Proc. Asian. Conf. of the 20th on Remote Sensing. Hong Kong, China, pp 939 –944 (1999).
- Tseng, D. A., and Huang, M. Y., Automatic
thresholding based on human visual perception. Image and Vision
Computing.,11:539-548 (1993).
|
|
|