SimpleStereo 1.0.7

Stereo vision made simple

SimpleStereo

Stereo vision made Simple.

SimpleStereo is a high level framework for stereo vision applications. It is written in Python 3, with C++ extensions. Documentation is available at https://decadenza.github.io/SimpleStereo/

Dependencies

• Python 3 (tested with 3.9.2)
• NumPy
• SciPy
• OpenCV
• matplotlib (for data visualisation)

Installation

Before starting, be sure to have the latest setuptools package by running pip install --upgrade setuptools. Then proceed with one of the two options below.

Option 1

Install package from PyPI with:

pip3 install simplestereo

Option 2

Clone or download the latest version and unzip. Then, from the folder containing setup.py, run:

pip3 install .

Troubleshooting

I am aware of some issues while installing SimpleStereo. If you have errors during installation, please open an issue.

Windows users troubleshooting

If during installation you get, together with other messages, the following error:

error: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.microsoft.com/visual-cpp-build-tools/

Please install the Microsoft C++ Build Tools as indicated. These are required to build C++ extensions that are part of SimpleStereo. More information about compiling on Windows is available at https://wiki.python.org/moin/WindowsCompilers.

Basic example

SimpleStereo helps you with common tasks. You can calibrate two cameras and initialise a stereoRig with:

import simplestereo as ss

# Path to your images
images = [
    ("0_left.png", "0_right"),
    ("1_left.png", "1_right"),
    ("2_left.png", "2_right"),
    ...
    ]

# Calibrate and build StereoRig object
rig = ss.calibration.chessboardStereo( images, chessboardSize=(7,6), squareSize=60.5 )

# Save rig object to file
rig.save("myRig.json")

# Optionally print some info
print("Reprojection error:", rig.reprojectionError)
print("Centers:", rig.getCenters())
print("Baseline:", rig.getBaseline())

More examples available in the examples folder.

Features and Future Work

General

• StereoRig, RectifiedStereoRig and StructuredLightRig classes to easily manage your stereo rigs and their calibration
• Basic stereo capture using OpenCV cv2.videoCapture
• Export and import point cloud to PLY file

Calibration algorithms

• Chessboard calibration (one and two cameras)
• Camera-projector calibration adapted (Moreno D. et al.), adapted from procam (ss.calibration.chessboardProCam)
• Camera-projector calibration alternative version (ss.calibration.chessboardProCamWhite)
• ArUco camera calibration algorithm

Stereo rectification algorithms

• Fusiello et al., A compact algorithm for rectification of stereo pairs, 2000 (ss.rectification.fusielloRectify)
• Wrapper of OpenCV algorithm (ss.rectification.stereoRectify)
• Loop and Zhang, Computing rectifying homographies for stereo vision, 1999 (ss.rectification.loopRectify)
• Lafiosca and Ceccaroni, Rectifying homographies for stereo vision: analytical solution for minimal distortion, 2022, https://doi.org/10.1007/978-3-031-10464-0_33 (ss.rectification.directRectify, see also DirectStereoRectification)

Passive stereo matching algorithms

• Adaptive Support Weight algorithm (K. Yoon et al., Adaptive support-weight approach for correspondence search, 2006)
• Geodesic Support Weight algorithm (simplified implementation, credits Asmaa Hosni et al.)

Active and Structured light algorithms

• Gray code
• De Brujin
• Adapt structured light algorithms to work with two cameras
• StereoFTP (Lafiosca P. et al., Automated Aircraft Dent Inspection via a Modified Fourier Transform Profilometry Algorithm, Sensors, 2022)

Unwrapping algorithms

• Infinite impulse response (Estrada et al., Noise robust linear dynamic system for phase unwrapping and smoothing, Optics Express, 2011)

Documentation

Documentation follows numpydoc style guide.

Install documentation prerequisites with:

pip install Sphinx numpydoc

Build documentation with:

cd sphinx-documentation-generator
sh BUILD_SCRIPT.sh
cd ..

Contributions

Reporting issues and proposing integrations of other stereo vision algorithms is highly encouraged and it will be acknowledged. Please share your issues!