A Cooperative Multi-Agent Approach for Aerial Image Rectification
Aerial images provide instant visual records which are closer to the areas of interest. As one of the extensively used methods of remote sensing, accurate aerial images could provide useful spatial information. However, aerial images are distorted during the process of acquisition by different reaso...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/26760/1/Liew%20Lee.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| id |
my-unimas-ir.26760 |
|---|---|
| record_format |
uketd_dc |
| institution |
Universiti Malaysia Sarawak |
| collection |
UNIMAS Institutional Repository |
| language |
English |
| topic |
QA75 Electronic computers Computer science |
| spellingShingle |
QA75 Electronic computers Computer science Liew, Lee Hung A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| description |
Aerial images provide instant visual records which are closer to the areas of interest. As one of the extensively used methods of remote sensing, accurate aerial images could provide useful spatial information. However, aerial images are distorted during the process of acquisition by different reasons such as earth curvature, camera lens and inconsistencies in the attitude of aircraft. Distortion is usually composited differently, collectively and irregularly in the entire aerial image which forms the complexity of degradation to the images. Hence, the aerial images should be rectified before proceeding to subsequent analysis. Image rectification is an essential pre-processing step to eliminate or at least reduce the effect of distortion. The conventional non-parametric approach requires a set of corresponding control points to be selected manually from a reference image and a distorted image as mapping parameters for rectification transformation. The procedure of manual selection is time-consuming and influenced by humans factors such as experience and attitude. Although many automatic extraction methods could be used conveniently to produce a large number of corresponding control points, excessive control points would also introduce other displacements into the image. Many trials usually have to be traversed to filter the original set of control points in order to select the best set of control points. Furthermore, the selected control points are advised to be distributed evenly or uniformly which is subjective to be defined. Such tedious process involves human, software and technology. Hence, this research introduces a cooperative multi-agent system approach to support the selection of control points for aerial image rectification. It incorporates the distribution of control points using Voronoi diagram and corner score through the cooperation and interaction of software agents. The proposed approach is tested on rectifying a two dimensional (2D) aerial image. The aerial image size used is 4000 × 2250 pixels. The area of interest covers a relatively flat area of almost 270,000 square meters. The coordinates of ground control points are retrieved from the Sarawak Land and Survey Department, which has an orthophoto system with 0.2 metres resolution. The performance of the rectification is measured using the total root mean square error of a set of checkpoints. The results obtained from the selection by the proposed multi-agent system are compared with results obtained from the selection by expert and conventional residual. The selection of ground control points is performed with the selection size of 3, 6, 9, 12, 15, 18, 21, 24 and 27. The selected ground control points based on the selection size are used in coefficients calculation for various transformations. From the finding, it is observed that the proposed multi-agent system achieved 0.91 meters as its lowest total RMSE of checkpoints when using the selected 18 ground control points and with the second order polynomial transformation. The proposed multi-agent selection achieved the highest percentage of reduction in the total RMSE of checkpoints with selection size of six and the second order polynomial transformation. The selection by the proposed multi-agent system achieved a decrement of the total RMSE of checkpoints with 58.31% when compared to the achievement by expert selection. The proposed multi-agent system selection also has a decrement of 12.41% total RMSE of checkpoints as the highest when compared to the achievement by the conventional residual method. Experiments show that the proposed multi-agent system provides reliable results. The study demonstrates that multi-agent approach is a way of representing task allocation, team planning and open environments which could support in providing a systematic way of structuring an aerial image rectification. |
| format |
Thesis |
| qualification_name |
Doctor of Philosophy (PhD.) |
| qualification_level |
Doctorate |
| author |
Liew, Lee Hung |
| author_facet |
Liew, Lee Hung |
| author_sort |
Liew, Lee Hung |
| title |
A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| title_short |
A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| title_full |
A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| title_fullStr |
A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| title_full_unstemmed |
A Cooperative Multi-Agent Approach for Aerial Image Rectification |
| title_sort |
cooperative multi-agent approach for aerial image rectification |
| granting_institution |
Universiti Malaysia Sarawak (UNIMAS) |
| granting_department |
Faculty of Computer Science and Information Technology |
| publishDate |
2019 |
| url |
http://ir.unimas.my/id/eprint/26760/1/Liew%20Lee.pdf |
| _version_ |
1783728325659721728 |
| spelling |
my-unimas-ir.267602023-06-26T03:50:51Z A Cooperative Multi-Agent Approach for Aerial Image Rectification 2019-09-10 Liew, Lee Hung QA75 Electronic computers. Computer science Aerial images provide instant visual records which are closer to the areas of interest. As one of the extensively used methods of remote sensing, accurate aerial images could provide useful spatial information. However, aerial images are distorted during the process of acquisition by different reasons such as earth curvature, camera lens and inconsistencies in the attitude of aircraft. Distortion is usually composited differently, collectively and irregularly in the entire aerial image which forms the complexity of degradation to the images. Hence, the aerial images should be rectified before proceeding to subsequent analysis. Image rectification is an essential pre-processing step to eliminate or at least reduce the effect of distortion. The conventional non-parametric approach requires a set of corresponding control points to be selected manually from a reference image and a distorted image as mapping parameters for rectification transformation. The procedure of manual selection is time-consuming and influenced by humans factors such as experience and attitude. Although many automatic extraction methods could be used conveniently to produce a large number of corresponding control points, excessive control points would also introduce other displacements into the image. Many trials usually have to be traversed to filter the original set of control points in order to select the best set of control points. Furthermore, the selected control points are advised to be distributed evenly or uniformly which is subjective to be defined. Such tedious process involves human, software and technology. Hence, this research introduces a cooperative multi-agent system approach to support the selection of control points for aerial image rectification. It incorporates the distribution of control points using Voronoi diagram and corner score through the cooperation and interaction of software agents. The proposed approach is tested on rectifying a two dimensional (2D) aerial image. The aerial image size used is 4000 × 2250 pixels. The area of interest covers a relatively flat area of almost 270,000 square meters. The coordinates of ground control points are retrieved from the Sarawak Land and Survey Department, which has an orthophoto system with 0.2 metres resolution. The performance of the rectification is measured using the total root mean square error of a set of checkpoints. The results obtained from the selection by the proposed multi-agent system are compared with results obtained from the selection by expert and conventional residual. The selection of ground control points is performed with the selection size of 3, 6, 9, 12, 15, 18, 21, 24 and 27. The selected ground control points based on the selection size are used in coefficients calculation for various transformations. From the finding, it is observed that the proposed multi-agent system achieved 0.91 meters as its lowest total RMSE of checkpoints when using the selected 18 ground control points and with the second order polynomial transformation. The proposed multi-agent selection achieved the highest percentage of reduction in the total RMSE of checkpoints with selection size of six and the second order polynomial transformation. The selection by the proposed multi-agent system achieved a decrement of the total RMSE of checkpoints with 58.31% when compared to the achievement by expert selection. The proposed multi-agent system selection also has a decrement of 12.41% total RMSE of checkpoints as the highest when compared to the achievement by the conventional residual method. Experiments show that the proposed multi-agent system provides reliable results. The study demonstrates that multi-agent approach is a way of representing task allocation, team planning and open environments which could support in providing a systematic way of structuring an aerial image rectification. Universiti Malaysia Sarawak (UNIMAS) 2019-09 Thesis http://ir.unimas.my/id/eprint/26760/ http://ir.unimas.my/id/eprint/26760/1/Liew%20Lee.pdf text en validuser phd doctoral Universiti Malaysia Sarawak (UNIMAS) Faculty of Computer Science and Information Technology |