Horizontal and Vertical planes on walls

[BBridger]

Hi. I am new to the forum. I am doing a small project comparing the spatial accuracy of laser scan data to terrestrial photogrammetry. My test subject is a wall, on which I have mounted three targets. I have a "known" distance between the targets and measuring the distances between the targets on the scan data and terrestrial photogrammetry. I keep running into an issue where the horizontal and vertical plane of the wall generated from the terrestrial photogrammetry (processed in 3DF Zephyr) is warped (i.e., not true). It appears as if the wall is following the path of the camera locations. I have attempted to keep the distance from the wall the same for each station, but even small variations in position seem to adversely affect the results. Has anyone in the forum come across this before? Any tips on collection process to avoid this?

[Andrea Alessi]

Hi BBridger,

it sounds like you're suffering from a bad calibration and/or drifting error. If the subject is not very textured, as a white wall with just 3 targets, that's a very strong possibility.

I suggest placing something very textured on the wall, such as newspapers sheets, which will help a lot with the orientation. Make sure you are taking images not just in a straight line, but also with a different depth. You can also use control points as constraints (even 2D only) to optimize your camera parameters, or as a last resort, use the optimize camera parameters using a terrain assumption since you're just using a plane.

I suggest doing your test on a very texture subject (as for example a brick wall) rather than a white wall, if this is the case. If you want to shar eyour dataset with me I will be happy to assist.

[BBridger]

Thanks for the response Andrea. With the last group of students I was working with we put some different colour targets up on the wall and that seemed to do the trick. Although the wall in question is partially brick, it is white, so adding the extra colour really helped.