Automatic guideline generation for parking assistance system based on on-screen display

The rear-view camera at the vehicle's backside is equipped for parking supports display system. It requires a wide-angle lens for obtaining a broad landscape. Throughout the wide-angel lens, an input picture is converted the Barrel distortion. Therefore, a driver cannot make out the distorted input picture in order to overlay the track directly. In this paper, we propose the generation automatic guideline based on on-screen display (OSD) that has new distortion using the intelligent neural networks. The neural network is to harmonize the distorted input picture with the generated guideline which is composed of the parameter of the car-width, the extension line, the expectation track, the camera position, the steering angle, the characteristic of lens, and so on. Experimental results show that the proposed method supports the driver for being able to read out the distorted input picture and the generated guideline.