Deep learning for ground penetrating radar image processing
This project focus on finding the relationship between 2D B-scan images and 3D B-scan images generated by gprMax for convolutional neural network training purpose in object detection and classification. Cylinders with different material and orientations have also been discussed to find the effects o...
Saved in:
| 主要作者: | |
|---|---|
| 其他作者: | |
| 格式: | Final Year Project |
| 語言: | English |
| 出版: |
Nanyang Technological University
2020
|
| 主題: | |
| 在線閱讀: | https://hdl.handle.net/10356/140154 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 機構: | Nanyang Technological University |
| 語言: | English |
| 總結: | This project focus on finding the relationship between 2D B-scan images and 3D B-scan images generated by gprMax for convolutional neural network training purpose in object detection and classification. Cylinders with different material and orientations have also been discussed to find the effects on the accuracy of 2D B-scan modelling. Results show that 2D B-scan images are very similar to that of 3D B-scan images of cylinders. In terms of shape and relative position of hyperbola, 2D B-scan images are exactly the same as 3D B-scan images. The minor difference in colour intensity can be negligible. The change in orientation of cylinders can also be represented in 2D B-scan images. |
|---|