OpenCV Python Epipolar Geometry Stereo Vision
ฝัง
- เผยแพร่เมื่อ 11 ก.ค. 2024
- In this video, I will go over epipolar geometry in OpenCV using Python in VS Code. I will show you how to find the fundamental matrix and use that to draw epipolar lines (epilines) in the two pair of stereo images from the Middlebury dataset.
0:00 Introduction
0:23 What is epipolar geometry?
2:59 Why do we need epipolar geometry?
3:19 How does epipolar geometry work?
5:28 Code
Thanks for watching! If you found this video helpful, please like, subscribe and share:
/ @kevinwoodrobotics
Social:
LinkedIn: / kevinwoodrobotics
Github: github.com/kevinwoodrobotics
Instagram: / kevinwoodrobotics
Twitter: / kevinwoodrobotics
Code and Doc: kevinwoodrobotics.com/product/opencv-python-epipolar-geometry-stereo-vision/
OpenCV Python Playlist Code and Doc: kevinwoodrobotics.com/product/opencv-python-tutorials-full-playlist/
Such a great video worth watching. Well explained
Glad you found it useful!
Thank you for sharing some information❣️💙
Thanks!
Amazing broo🎉
awesome
this will be cool to have realtime in some compute shaders 🙂
Yeah definitely
Great video!!
Thanks!
👩🏻💻✨✨✨
❤️
Thanks for sharing this video, Kevin!
I have a question and hope you could help me understand it. At 9:50, regarding the function cv.computeCorrespondEpilines, could you explain the parameter 'whichImage' in this case in terms of left or right image? I wasn't so sure if 2, the second image, was the left or the right one. If we have keypoints of the right image, should we we draw the epipolar lines on the left image? So, in this case, 2 is the left image. Am I correct? Thanks a lot Kevin!
Never mind. I went to the documentation for C++ and got it now. It's the image that contains the points. I guess the documentation for Python is a bit confusing for me since it says whichImage 'specifies for which image epilines need to be drawn'.
Very good question! This can be confusing, but the main concept here is that a point that appears to be in one image is actually a line in the other image. The F matrix is used to exploit that relationship. So left is actually image 1 and we use the points from image 2 (right) to draw the lines in the left image and vice verse. Hope this helps!
@@kevinwoodrobotics Yes! it helps a lot. Thanks, Kevin.
you used 2 photos with a parallel angle, but will this code work in the second case, when the photos are at angles? I want to try it out. I really liked your explanation
Yes it should still work. It will just change the slope of your elipolar lines
1st tuning
😁