Texture mapping: occlusion logic of faces when projected to camera

• PCL Version: commit dd532fe

I think there is a logic problem in texture_mapping.hpp.
code_link

if (idx_pcam == current_cam && !visibility[idx_face])
{
         // we are now checking for self occlusions within the current faces
         // the current face was already declared as occluded.
         // therefore, it cannot occlude another face anymore => we skip it
         continue;
}

We want to project all faces to a camera plane and determine which face is occluded and which one is visible.

If face A occlude face B, face B occlude face C.
Imaging we check face A first, find out that B is occluded by A, B is marked non-visible.
And then we check B, since B is non-visible, we skip.
At the end, We conclude that no faces occlude C, so C is visible, which is not true.

I don't know if i had missed something or maybe I just didn't fully understand the code.
Can someone figure it out for me? Thx.

Start in surface/include/pcl/surface/impl/texture_mapping.hpp around line 851 and trace how projected faces update visibility during self-occlusion checks. Reproduce the reported A→B→C chain, then verify that the occlusion logic does not skip a face in a way that incorrectly leaves C visible.