Automatic calibration algorithm that calibrates four cameras of an around view monitor (AVM) system simultaneously in a natural driving situation
-
Features and Process:
- Converts fisheye images of the surrounding views of a vehicle into bird-eye images
- Extracts lanemarks on a street in a driving environment
- Remove falsely detected lanemarks based on orientations and sorting vanishing points using a voting box algorithm
- Calibrate the cameras using Levenberg–Marquardt (LM) algorithm
-
Libraries and APIs used: C , Ceres Solver, Eigen, OpenCV
-
This project is an attempt to realized the paper Automatic Calibration of an Around View Monitor System Exploiting Lane Markings by Kyoungtaek Choi, Ho Gi Jung, and Jae Kyu Suhr