How to calculate real-world distance from object mask and centroid (D405)?

[qidihan]

---

Required Info	How to calculate real-world distance from object mask and centroid (D405)?
Camera Model	{ D405 }
Firmware Version	(The newest)
Operating System & Version	{ Linux 22.04
Kernel Version (Linux Only)	()
Platform	PC
SDK Version	{ The newest }
Language	{Python }
Segment	{Robot }

Issue Description

Hello,

I am working on a computer vision project where I need to calculate the real-world distance that an object has moved based on its mask size and centroid coordinates in the image. I have the following information available:

I can extract the object mask (binary segmentation) from the image. I have the centroid coordinates (x, y) of the object within the image. I use 640*360 resolution. However, I am unsure how to convert these image-based measurements into real-world distances (e.g., in millimeters or meters). Can you help me on this?

Thanks, Qidi

[MartyG-RealSense]

Hi @qidihan If you have XY image pixel coordinates then you can convert those coordinates to real-world distance in meters by converting an RGB color pixel to a depth pixel with the instruction rs2_project_color_pixel_to_depth_pixel. Python examples of implementing this instruction can be found at https://github.com/IntelRealSense/librealsense/issues/5603#issuecomment-574019008 and https://github.com/IntelRealSense/librealsense/issues/9945#issuecomment-972539640

[qidihan]

I may not have been clear. What I meant was that I am using a 640*360 RGB image, and I want to know approximately how many millimeters a pixel is in this image, not the distance of my object from the camera. I found online that millimeters can be calculated as pixels multiplied by (25.4 / PPI), but I don't know the PPI value.

[MartyG-RealSense]

There is not RealSense-specific information available about this. My general non-RealSense research of the subject indicates though that in order to calculate PPI, you need to know the diagonal size of the screen that you are using.

https://www.calculatorsoup.com/calculators/technology/ppi-calculator.php

The DPI (Dots Per Inch) value is knowable though. At 640x360 resolution, DPI should be 160.

The guide below asserts in the section headed "Are PPI & DPI the same?" that PPI and DPI can be used interchangeably (1 PPI = 1 DPI), so that 160 DPI would be equal to 160 PPI. Other resources back up the assertion that DPI and PPI are interchangeable.

https://cssunitconverter.com/pixels-to-inches/

[qidihan]

Thanks for your information!!! I will close the issue right now.

[MartyG-RealSense]

You are very welcome. I'm pleased that I could help. Thanks very much for the update!