Bookmark and Share

Hardware Topics

Making Stereo Camera
1 reply [Last post]
Last seen: 7 years 12 weeks ago
Level 1: Prestidigitator
Joined: 2012-10-31
Points: 4



Is it difficult to make your own stereo camera? 

I found this article

It seems simple, but the article doesn't go over camera synchronisation, is this important? If images were not synchronised would this lead to problems?

Also how difficult is camera calibration, i know you would have to calibrate for lens distortion and camera alignment. 

I am involved in a project, we are considering using a stereo camera to detect small rocks & craters in a very flat and dusty environment.

Amit Shoham
Last seen: 20 weeks 1 day ago
Level 4: Thaumaturgist
Joined: 2011-05-28
Points: 60

The answers to your questions largely depend on the details of what you're trying to accomplish.

For example, synchronization is critical if there's substantial motion in the video (either due to camera motion/vibration or object motion).  What constitutes "substantial" motion?  Well, that depends on both the frame rate you're using and the precision required in the results.  If you know that your cameras and the terrain will both be perfectly still, then there's no need for synchronization.  If you expect lots of moving, shaking, etc. then you'll absolutely need to synchronize.

Similarly, the need for alignment and calibration might vary depending on your application.  If you need to measure the exact size and/or position of small rocks and craters, then you'll want to use fairly traditional techniques which require careful calibration.  If you need very precise results, expect this to be difficult. 

On the other hand, I know that some companies are developing stereo algorithms that adaptively calibrate "on the fly" much like the human brain.  You wouldn't need to do any calibration at all, but--like the human visual system--you would get only rough relative data.  I should add that I haven't seen any open source for such algorithms, so there may be some R&D involved.