I’m having trouble getting a good bottom of curb elevation when generating the DSM and then using Virtual Surveyor to extract the data. The face of curb is not well defined like a wall or building. This is a problem for processing road corridors.
You can see below how the drop inlet is pulling into the curb line and creating a u shape at the bottom.
Is this a flight issue or a processing issue.
DSM was created with multiple smoothing and non smoothing settings.
This is “good enough” to manually digitizing in the point cloud, but not clean enough to automate. Another method would be digitizing the top-back curb and then offsetting & lowering the polyline by values measured in the field to pick up the flowline. Again, not reliable for design surfaces. Carlson Point Cloud has some pretty cool supervised digitizing options for curbs in the point cloud, but I’ve found it unusable on anything other than a LiDAR point cloud. We typically do hard break lines in the field, if it’s for design.
Anyways, subscribed to this post & looking forward to seeing the followup from Pix4d.
The OneDrive link does not work. Is there another way to add the Quality Report?
My flight was done at 200’ AGL using Pix4Dcapture and capturing both Nadir and Oblique Photos. The total project length was roughly 3000’ have a total of 763 images.
Derrick - Interesting… I typically locate some of the hard break lines and structures as well. I’m not fully comfortable yet with using this tech for design purposes. Although I have seen some very good checks from field collected data to my aerial data. But seeing the drop inlet pull into the curbline makes me a bit uneasy.
Zach - thank you for the files. I checked the Quality Report and your other outputs.
I am aligned with Derrick and his suggestions about the flight parameters and digitizing method that he described in the post above.
In general, what we can say about improving the quality of road reconstruction is to try adjusting some settings in the software by:
adding manual tie points or Ground Control Points (GCP)
it will help in the process of matching the images together, and it can result in better-reconstructed scenes,
Looking at the point cloud of the road what worries me is the fact that the street is still not smooth enough, even after multiple smoothing settings you applied. The difference in distance between the top and the bottom point in the profile of the road seems to be too high.
What could improve the results would be:
the change in the minimum number of matches from 3 to 4 or 5 (already described above)
but also unchecking the multiscale option. In some cases, this move dramatically improves the quality of the outputs.
Just have a look at the tower example:
In my opinion, that could also help you smoothing the road and obtain better subsequent outputs.
Please let me know what you think guys. We can discuss the topic further, sharing knowledge, trying to find more possibilities and solutions.
Multiscale should defintely be disabled on road projects, in my experience. My opinion is that if the points can’t be built with confidence, they shouldn’t be built at all. I would much rather see the gaps in areas with low visual content and be able to identify them, as opposed to having to manually clean up noise around points that are low confidence anyways. The best luck I’ve had with curbs are using the flight settings I listed above (“low & slow”), and also high overlap for the sole reason that I want to increase my number of matches to 4 or 5, as Beata said. Also, don’t even bother with all of this if you’re doing an as-built on a brand new road. Fresh asphalt is a no-go.
There’s a lot of areas where drones, photogrammetry & pix4d really come in handy and provide a benefit to us and our clients. Design surveys in dense urban areas are one of those if you’re talking 2D, but the 3D data can be very hit and miss. I’d guess that ~75% of the time, we only use the drone to produce an orthophoto in those situations and just have the point cloud ready to extract features that may have been missed in by the field crew, like fire hydrants or powerpoles. They also serve as a great check/red flag for bad field shots, and the orthos are very low-effort to produce, as you know. We tend to save the digitizing & surface building drone uses for projects where we know it’s effective and efficient, like rural topos, landfill calcs and grading analysis.
Thanks for the help. I will reprocess with your suggestions and report back.
Is there a negative effect of adding too many GCPs or Manual Tie Points? I used 14 on this project but I could have added MTP’s at additional locations that were field checked I suppose.
These cookies are necessary for the website to function and cannot be switched off in our systems.
They are usually only set in response to actions made by you which amount to a request for services, such as setting your privacy preferences,
logging in, or filling in forms. These cookies do not store any personally identifiable information.
These cookies allow us to count visits and traffic sources so we can measure and improve the performance of our site.
They help us to know which pages are the most and least popular and see how visitors move around the site.
All information these cookies collect is aggregated and therefore anonymous.
If you do not allow these cookies we will not know when you have visited our site, and will not be able to monitor its performance.
These cookies may be set through our site by our advertising partner (Google).
They may be used by Google to build a profile of your interests and show you relevant adverts on other sites.
They do not directly store personal information but are based on uniquely identifying your browser and internet device.
If you do not allow these cookies, you will experience less targeted advertising.