Sensor transformation is wrong for registered PTX files

[framed]

Hi Daniel,

Pretty sure this is a bug. When importing a PTX file that has been registered (eg. exported from Leica Register360), the Sensor coordinates come through at 0,0,0. This is incorrect. The sensor coordinates should be translated by the values in row 10. See here: https://wiki.photoneo.com/index.php?tit ... ile_format

Furthermore, lines 7-10 define the transformation matrix for the sensor. See screenshot below, which shows the transformation matrix in a registered PTX next to the correct sensor location in CC (from an E57 file of the same cloud)

[daniel]

Which version are you using?

[framed]

Hi Daniel. Have tried on 2.13.2 and 2.14.beta. Behaviour is the same on both.

[daniel]

So referring to the code and to https://paulbourke.net/dataformats/ptx/, the sensor position and orientation is stored in rows 3-6 (row 3 is the sensor center, and the next 3 rows are the X, Y and Z vectors).

Rows 7-10 store the global transformation matrix that applies to the points (and I would assume to the sensor as well). At least that's what I could see in samples files such as some files found on http://www.libe57.org/data.html or Paul Bourke's site.

ptx_header.jpg

So in your case, I would expect row 3 to equal the 3 first values of row 10, and rows 4 to 6 equal the 3 first values of rows 7 to 9 (or something totally different depending on how the cloud was registered). Otherwise, this means that the points have been registered, but there's no 'sensor' associated.