Photo to las - Beginner question

[Berke Z.]

Hi everyone,

I'm working on a project that includes a panoramic camera + Velodyne LIDAR for street imaging. I developed an image processing app to detect specific objects in pictures but need to get the measurement from LIDAR. 

I have Las files from the LIDAR scene and a folder of images from street imaging. How can I get real-world measurements from lidar objects within the specified coordinates from panoramic photos? Or is this even possible?

I'm a newbie in this field so If my question could be abstract sorry in advance. Any help would be amazing for me thanks.

Thanks for reading

[Jochen Rapidlasso]

Hi Kayit Berke,

not sure if I got your question. LAS/LAZ stores points, not objects. Every Point has its coordinates (X/Y/Z) in space. So its quite easy using simple math to get the measurement like distance of points to each other. (The more complex part is may more to detect an object and may get the outer boundaries of an object.) 

So tell more about your problem, best by a small sample and nice pictures :)

Best regards,

Jochen @rapidlasso

[Berke Z.]

Hi  Jochen,

Thanks for answering. I have a Leica Mobile Mapping system; GPS, LIDAR, and panorama photos of the streets. I detected signboards in the images with artificial intelligence. I have image coordinates of every point but what I don't have corresponds to these coordinates on point cloud data. The Leica Manager software provides me this but manually with matching images and Point Cloud data.  Nonetheless, I was wondering how can I match these images and point cloud data so I can measure real-world signboards' areas automatically. Or is this even possible with python?

I really would like to share some photos but due to GPDR law, I possibly cannot. For better understanding, you may think Google Maps Street View and see the signboards and Point Cloud data behind them.

Thanks again for responding, have a lovely day.

Berke

20 Mart 2023 Pazartesi tarihinde saat 15:00:58 UTC+3 itibarıyla Jochen Rapidlasso şunları yazdı:

[Jorge Delgado García]

Of course, it is possible, but you need to know each image what are its geometrical model, and its own orientation parameters (internal and external). These expressions are the ones that allow you to relate the image space (pixels measured in the image) with the terrain space, or vice versa (terrain coordinates, for example, LiDAR points with the coordinates in which this point appears in each of the images you have).
For conventional cameras you can use the collinearity model (together with the internal camera orientation or self-calibration parameters), for other types of cameras (fisheye, panoramic, etc.) you must use the specific models.
If you know the orientation parameters of the images (and their geometrical model) and the object appears at least in the two images, everything is reduced to an intersection of rays, starting from the projection centers of each image and passing through these points (homologous points), the intersection will be the ground point XYZ.

If you need more detail do not hesitate to contact me.

Jorge

jdel...@ujaen.es

--
Download LAStools at
https://rapidlasso.de
Manage your settings at
https://groups.google.com/g/lastools/membership
---
You received this message because you are subscribed to the Google Groups "LAStools - efficient tools for LiDAR processing" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lastools+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lastools/72ec338b-61aa-47c0-a426-84fa1d92cec7n%40googlegroups.com.

--