stereo-calibration-and-3D-reconstruction

This developed project contains files ...

• to capture calibration images / test images using stereo cameras
• calibrate cameras using captured calibration imags
• Rectify the test images using saved calibration parameters
• Calculate depth at the pixel of the image/live video feed using mouse click

The chessboard pattern pattern.png is included in the repository. or download it from:

https://github.com/opencv/opencv/blob/master/doc/pattern.png

Requirements

Intstall below packages using pip.

• Two webcams or stereo camera
• numpy==1.16.6
• opencv-python==4.4.0.46
• matplotlib

Steps

1. Print the chessboard pattern and paste it on a hard surface

2. Fabricate stereo camera pair and fix it on a hard board, so it is fixed and does not move while using

3. Capture multiple images of the chessboard pattern from the different views for calibration using CaptureCalibrationImages.py

4. After capturing sufficient amount of images, calibrate stereo camera pair using CalibrateStereocamera.py

5. To check the efficiency of the calibration, click some test images using CaptureTestImages.py and check the rectification efficiency using Rectification.py.

6. Use displayDepth_image.py or displayDepth_video.py to compute disparity and depth maps. The depth values are calculated at the pixel location with the mouse click.