calibrated principal distance vs focal length for 160FOV fisheye

OpenCV, Python and other image processing questions
Post Reply
onestarhalfpoint
Posts: 1
Joined: Mon Jun 06, 2022 4:35 am

calibrated principal distance vs focal length for 160FOV fisheye

Post by onestarhalfpoint »

Hello,
I have tried using the stereopi V2 setup with two 8MP (IMX219) 160FOV waveshare cameras, to shoot several pairs of still photos at max resolution (6560x2464). I then started my calibration following your github code (the latest fisheye_robot one). I managed to get intrinsic matrix, K, for both left and right cameras, from which I found the "principal distance" at the diagonal terms, which are around 1575 pixels. If I googled it correctly, the IMX219 sensor has a pixel size = 1.12um = 1.12E-3 mm. Therefore, the principal distance in units of mm is 1575 x1.12E-3 =1.764mm. Here goes my question! per the waveshare website, https://www.waveshare.com/imx219-d160.htm, the 160FOV camera has a specified focal length 3.15mm, much higher than my calibrated principal distance of 1.76ish mm. I understood there is a conceptual difference between focal length and calibrated principal distance, but expected to be close. I am wondering has anyone checked the calibrated principal distance for stereopi v2 fisheye setup, e.g., vs the focal length specified at official website? The example pictures on your github were not taken at its max resolution, I am not sure how should we directly use it to check focal length? Many thanks for any help!

User avatar
Realizator
Site Admin
Posts: 900
Joined: Tue Apr 16, 2019 9:23 am
Contact:

Re: calibrated principal distance vs focal length for 160FOV fisheye

Post by Realizator »

Hi onestarhalfpoint,

I trust the OpenCV calculations but not the specifications declared on the website. :-) To say more, you can find a few similar camera models from different manufacturers with dramatically different focal length numbers.
Eugene a.k.a. Realizator

Post Reply