If I want to monitor roads condition(whether they are in good condition or any need to repair), how can I do this using satellite data. If possible, then how one can perform the same in Python or any other programming language.
After searching, I got to know few things regarding with this question. Sentinel satellite one pixel is of 10 m, whereas roads are not that much wide. If we consider this data then there is possibility of mixed pixels. So we can apply Learning & Model development approach. One can compare village roads & highway properties to develop a model for road health monitoring. We can analyse differences across various features.
Can someone please add more thoughts in this one?
And what about using synthetic aperture radar (SAR) data ? You can get data with spatial resolution of a meter or so, and if you want to monitor surface movement or detect changes, it is certainly highly adequate. Have a look at the link below (other sensors exist).
https://en.wikipedia.org/wiki/TerraSAR-X
Directly recognising damage to the road will be difficult. But there will be information about driver behaviour: are there queues in unexpected places, e.g. not at a junction?
However, there may be better ways to get this information, e.g. tracking mobile phones to identify locations where drivers unexpectedly slow down.
If we have prior knowledge of the material used for construction of road like type of cement, it's spectral properties etc, we can train a model based on these properties as well as geometric properties of road. Then we can come with some solution.
Any suggestion??
Have you looked into hyperspectral imaging? There are satellites that take images of the earth across the spectrum to get more information. For each pixel, you get a vector of wavelength values. You end up with an image cube. It takes a lot of processing power, but modern processors can handle, especially if you use a GPU.
I worked on a project that analyzed surface condition using hyperspectral image processing. I don't know if the resolution is high enough for road conditions, but it sounds like the same problem.
One way could be to locate the scenes from the time close to a precipitation event. One strong determinant of road quality is water logging and drainage (especially true for India). Since water is relatively easy to see in both optical and radar imaging and is expected to have a significant effect on the signal strength. You would be in a position to comment on the road quality.
One way can be tracking the trend of the traffic flow. Like, the path not followed by most of the vehicles, Unnecessary and sudden speed change of vehicles in specific locations of the road. This technique can help in predicting the damaged locations in the road but directly recognizing damage to the road will be a bit tricky.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python