xmba15 / uncalibrated_stereo_rectification

simple library for uncalibrated stereo rectification using feature points

📝 uncalibrated stereo rectification

---

• original pair of images

• rectified images by method implemented in opencv (Hartley et al.'s method)

• rectified images by loop&zhang's method:

could not find Loop et al.'s original implementation so I implemented my own way of additional uniform scaling and translation after rectification.

🎉 TODO

---

• Implementation of uncalibrated rectification method presented in Computing Rectifying Homographies for Stereo Vision
• Implementation of DSR method

🎛 Dependencies

---

python setup.py install --record files.txt

• conda environment

conda env create --file environment.yml
conda activate stereo_rectification

🏃 How to Run

---

python scripts/test_stereo_rectification.py --left_image_path LEFT_IMAGE_PATH --right_image_path RIGHT_IMAGE_PATH

# eg: python scripts/test_stereo_rectification.py --left_image_path ./data/TOMMY2_L.png --right_image_path ./data/TOMMY2_R.png

the pair of images used here are from MCL SS Database

💎 References

---

• Theory and Practice of Projective Rectification
• Computing Rectifying Homographies for Stereo Vision, CVPR 1999
• DSR: Direct Self-rectification for Uncalibrated Dual-lens Cameras