Around view system is a system which enables the driver to see surrounding around the vehicle in such a way that it is being viewed from the top. Cameras placed around the vehicle capture images at an angle with respect to the ground normal. If the angle were parallel to the ground normal then the captured images look like being taken from the top. But it is too risky to fit cameras in that angle. Hence the images captured at an angle with the ground normal are processed to create a top view or bird’s eye view and then each of these images are registered for stitching. This facilitates displaying the surroundings of the vehicle in a single image and easier to analyse the safety during parking. Blind spots are eliminated and the driver can see the areas where he could not have seen before. The stitching is enhanced by blending using feathering to enhance the result. When the vehicle moves there is a chance for the camera to change its relative poses. Hence lane detection system is developed and around view was created using lanes. This automated calibration using lanes can update the extrinsic parameters when the relative poses of the cameras varies. Also when the positions of the lanes are known, the car can be provided with automated navigation. Vanishing point was also detected for creating a top view. Initial calibration and LUT creation was carried out with the help of a medium sized checkerboard which makes the system portable. After initial calibration of the multiple cameras, the images captured are processed and transformed using the initial calibration parameters stored as LUT.
