Class 4 of 4 on Multi-View Geometry covered in RRC Summer School 2022. The following topics are included here.
- Stereo camera depth estimation
- Triangulation
- PnP Formulation
The contents of this repository are as follows
| S. No. | Item Name | Description |
|---|---|---|
| 1a | stereo_to_depth.ipynb | Convert stereo rectified images to point clouds |
| 1b | stereo_to_depth_tsukuba.ipynb | Same thing as 1a, but with tsukuba (see difference between BFMatcher) |
| 2 | triangulation.ipynb | Triangulation of a single point |
The repository also contains the following
| S. No. | Item Name | Description |
|---|---|---|
| A | Slides - MVG4.pptx | Presentation |
- RRC Summer School 2022
- Schedule: laksh-nanwani.notion.site
- Stereo camera depth estimation
- Start with code: intelrealsense.com
- ML approaches: learnopencv.com
- Conceptual overview: medium.com
- C++ implementation with GUI controls: learnopencv.com
- Results and review: adept.net.au
- Triangulation
- PnP Algorithm
- PnP on Wikipedia
- PnP overview on OpenCV docs
- P3P
- Earliest works (1999): Linear N-Point Camera Pose Determination
- Recent solution (2003): Complete solution classification for the perspective-three-point problem
- Faster and more accurate (2018): Lambda Twist: An Accurate Fast Robust Perspective Three Point (P3P) Solver
- PnP
- Efficient O(n) PnP (2009): EPnP: An Accurate O(n) Solution to the PnP Problem
- SQPnP - non-polynomial solver which casts PnP as non-linear quadratic problem - (2020): A Consistently Fast and Globally Optimal Solution to the Perspective-n-Point Problem
- Other underlying theory
- Understand camera intrinsic matrix: towardsdatascience.com
