How to get intensity and elongation values?

[irmuun20]

Hello, I am new to lidar and want to test centerpoint 3D object detection model on data collected via ouster lidar. An ouster lidar returns following:

• Range - Range in millimeters, discretized to the nearest 1 millimeters with a maximum range of 524m. Note that range value will be set to 0 if out of range or if no detection is made.
• Calibrated Reflectivity - Sensor Signal Photons measurements are scaled based on measured range and sensor sensitivity at that range, providing an indication of target reflectivity.
• Signal Photons - Signal intensity photons in the signal return measurement are reported.
• Near Infrared Photons - NIR photons related to natural environmental illumination are reported.

The model, I am using, was trained on Waymo dataset which consists of following: x, y, z, intensity, elongation

• Intensity - A measurement indicating the return strength of the laser pulse that generated the LiDAR point, partly based on the reflectivity of the object struck by the laser pulse.
• Elongation - The elongation of the laser pulse beyond- its nominal width. Elongation in conjunction with intensity is useful for classifying spurious objects, such as dust, fog, rain. Our experiments suggest that a highly elongated low-intensity return is a strong indicator for a spurious object, while low intensity alone is not a sufficient signal.

I can produce x, y, z coordinates from Range value but I don't know how to get intensity and elongation from ouster lidar. I am not sure if Signal Photons from ouster lidar is the same as intensity in Waymo dataset.

I'd really appreciate it if someone will help me to get these two values from ouster lidar. Thanks in advance.

[kairenw]

Hi @irmuun20,

Do you know the units of the intensity measurement from Waymo? This issue here seems to indicate that it's raw photons? Is that true?

For elongation, I don't believe there is anything equivalent that we report from our sensor.

[irmuun20]

Yes it is raw intensity value that can go up to 6000

[kairenw]

Great - I'd suggest using SIGNAL for the intensity, though you may find you want to scale it to be able to use your trained model.

[irmuun20]

Yes that is the problem. when I scale it like following, it shows misaligned results: intensity = tanh(SIGNAL / 255.0) I wonder what is the correct way to normalize it.

[irmuun20]

When I use tanh(SIGNAL), the minimum signal(intensity) value in all points is 0.99.

[irmuun20]

https://github.com/HaohaoNJU/CenterPoint/issues/29#issuecomment-1461186498