Photo and Point Cloud Match Problem

Hello, I'm Ahn Soonhye uk, working in a company in Korea

I'm leaving a message because of the matching issue between the photo and the point cloud

I work with drones and mirrorless Leica scanners

However, the position of the drone and the position of the scanner are separated and matched due to the level problem

Scan data is e57 and merged data with noise and region already erased

Give them a hand.

Parents
  • You have not input any control points why should they match? Are the photos with accurate RTK positions? Have you added local geoid?

    If you do not want to manually input control points then you have to use two step process. At first align photos and then do another aerotriangulation with option to adjust photos to point cloud.

  • Thank you so much for the response from the Bentley community influencer

    1. You have not input any control points why should they match?

         (Is there a way to enter control points for pictures and scans? I know how to take pictures and control points)

    2. Are the photos with accurate RTK positions?

         (Dji drones have their own gps coordinates But there's no RKT signal from the mobile station)

    3. Have you added local geoid?

         (wgs84,ecef,korea2000/east belt 2010 (epsg:5187) coordinates alternately)

    4. If you do not want to manually input control points then you have to use two step process. At first align photos and then do another aerotriangulation with option to adjust photos to point cloud.

         

         (Could you tell me more about the two-step process? After matching the photos, I added cloud points and matched them, but I don't know what the option is to adjust the photos to fit the cloud you're talking about)

    Thank you for your kindness.

  • (Is there a way to enter control points for pictures and scans? I know how to take pictures and control points)

    No, GCPs are only for pictures. Scans should be prepared in correct coordinates.

    Could you tell me more about the two-step process?

    You have to run aerotriangulation with option Adjustment constraints: Point clouds (scans).

  • It's an honor to hear from you again

    1. Scans should be prepared in correct coordinates.

         (gps via dji drone)   (Gps of scan data entered by merging scan data via register 360 and exporting it to e57)

         Aren't the two correct coordinates? I'm curious.

    2. You have to run aerotriangulation with option Adjustment constraints: Point clouds (scans).

         As you said, I made a tutorial video Could you take a look and point out the wrong options or methods?

    I can't upload the video

    The video was not uploaded, so I uploaded it on my personal test homepage

    https://an-soon-hyeok.com/for-oto/

    There's no virus :)

    Thank you for your kindness

  • Aren't the two correct coordinates? I'm curious.

    Coordinate systems from GPS are the same on WGS84 but accuracy is not. Without RTK the DJI drone has consumer grade GNSS with accuracy in meters and vertical accuracy is relative(doesn't use GPS at all but takes altitude from barometer). Scanner is more accurate because you set scanner on known location.

    That is why you need to align them using GCPs or align with aerotriangulation

    As you said, I made a tutorial video Could you take a look and point out the wrong options or methods?

    None of your data has real coordinates as I see from video. Photos are from camera without GPS and point clouds are in local coordinates. It even warns you in 3D view that coordinate mismatch is detected.

    If they are not set in same coordinates you need to create so called fake GCPs and use that. Just read some coordinates from point clouds and use that as Survey points for photos then it will be in same place and can be used for reconstruction. Alignment step adjusting photos to point clouds then won't be needed.

  • Hello. Thanks to Mr. Oto, it's getting better

    1. Coordinate systems ae the same but accuracy is not. Without RTK the DJI drone has consumer grade GNSS with accuracy in meters and vertical accuracy is relative(doesn't use GPS at all but takes altitude from barometer). Scanner is more accurate because you set scanner on known location.

    That is why you need to align them using GCPs or align with aerotriangulation

         (I see. I understand for sure that the GPS of the drone is not accurate
          But I know how to put coordinates in a picture using gcp, but I don't know                  how to arrange using gcp
          Could you show me a link or a video of the method?
          I'm sorry for asking so many questions)

    2.  None of your data has real coordinates as I see from video. Photos are from camera without GPS and point clouds are in local coordinates. It even warns you in 3D view that coordinate mismatch is detected.

    If they are not set in same scoordinates you need to create so called fake GCPs and use that. Just read some coordinates from point clouds and use that as Survey points for photos then it will be in same place and can be used for reconstruction. Alignment step adjusting photos to point clouds then won't be needed.

         (Fake gps. I saw it in another community post
          I've read it about five times, but I didn't understand it.
          Can you explain it in detail or record it?
          (How to match photo data without gps data and scan data with fake gps.)
          If you can't upload video data due to capacity,
          Can you send it to ansoonhyeok331@gmail.com)

          You are my savior

  • Not fake GPS but fake GCP(ground control points).

    You simply read any coordinate from point cloud and then use it as Survey point(control point) and input this coordinate and mark it in at least 3 photos. You need to read coordinates from at least 3 points.

Reply Children