Synthetic LiDAR 3DPCD to Range image from .las input

So im applying a Hann window to a 2d range image of .las point cloud. Current window overlapping is 50% could this result in smearing the range image? see 3d point cloud and resulting range image below, my only other thought is that my projection may still be incorrect.

Answer 1

[Abhipsa]

Hello @Ryan,

The smearing is most likely due to the Hann windowing rather than the projection itself.

When you apply a 2D Hann window with 50 % overlap on a range image, neighboring windows blend overlapping pixels, which smooths out edges and creates a “smearing” effect — especially around depth discontinuities.

To verify:

1. Plot the raw range image (before the window) and confirm that the geometry looks correct relative to the original point cloud — if so, the projection is fine.
2. Try reducing the overlap (e.g., 0 % or 25 %) or using a smaller window size.
3. Alternatively, use a Gaussian or median filter with a limited kernel size instead of an overlapping Hann window if you only need light smoothing.

If you still see distortion even without any windowing, then the issue could be with how the LAS points are projected — for instance, the azimuth/elevation mapping or coordinate origin.

I hope this helps you.

[Ryan]

Thank you Abhipsa