| Title | The technology Development for Pavement Surface Distress Analysis |
| Dept | Transportation Engineering, Maritime and Air Transport Division |
| Year | 2010 |
| Month | |
| Price | |
| Summary | The pavement management agents used traditional pavement damage survey. It wasted a lot of time to judgment what is the needs to repair. So develop an automatic identification method in pavement damage survey is necessary. It used line-scan camera and wide-angle lens to capture 4m-width image by lane before identification. To overcome the harmful impact of sunlight and ambient light, the van used LED to balance the impact from surrounding light. After testing in site, it has a perfect image capturing result that the van speed under 60km/h. After light balance process, there still had kinds of pavement damage to the image. So combining the support vector machine and data mining method to determine a classify tree with a universal crack index. The accuracy of this method was over 90 % of the crack indentified. It also used the image feature analysis and grayscale standard deviation to judge the indention of manhole, pothole and patch. The accuracy of this method was over 80 %. The time to analysis one image needs 5-10 seconds, so this technology is efficiently in the pavement damage survey. It also provides a prototype to integrate the pavement management system and image database improving pavement management efficiently. Some faults in this system needs to improve and modify. For example, the patch area over then one image was difficult to indentify. So it also combines another equipment and technology to solve the problem in future. But this study is already suitable for pavement damage survey. |
| Post date | 2010/10/15 |
| Hashtags |
View count:
127