code-locker / stereo-vision

Stereo vision application using two cameras for depth and disparity calculation and volume estimation

Stereo-Vision

Abdullah_Zaiter__Ian_Moura
├── README.md (This file)
├── Abdullah_Zaiter__Ian_Moura.pdf
├── /data
│ ├── /Middlebury
│ │ ├── /Jadeplant-perfect
│ │ │ ├── imgL.png
│ │ │ ├── imgR.png
│ │ │ │── disparity.pgm
│ │ │ └── depth.png
│ │ └── /Motorcycle-perfect
│ │ │ ├── imgL.png
│ │ │ ├── imgR.png
│ │ │ ├── disparity.pgm
│ │ │ └── depth.png
│ └── /FurukawaPonce
├── /relatorio
│ └── Latex source code of the report
├── /src
| └── main.py (principal source code)
| └── utils.py (aux code with several relevant classes and methods)
| └── MouseClick.py (a class for mouse tracking)

OpenCV version : 3.4.1

Python 3

Python modules used:

 - cv2  
 - numpy  
 - argparse

Requisites:

  1- Calculation of disparity and depth map, and then using them to determine real world coordenates, this can be applied for two group of images, motor and plant.

  2- Calculation of disparity and depth for another group of image where there is a rotation between the two images.

  3- Determing the minimum box dimensions in which the object of requisite 2 can fit, this can be done clicking on the image two times to measure each dimension.

To evaluate requisite 1 run this command, where i parameter can be m or p (m for the motor images and p for the plant images):

python3 ./src/main.py --r 1 --i m OR python3 ./src/main.py --r 1 --i p

To evaluate requisite 2 run this command:

python3 ./src/main.py --r 2

To evaluate requisite 3 run this command (measuring box size):

python3 ./src/main.py --r 3

Photo of the result of the depth calculated