Modelling and Verification of Car Rear View Camera Using Ray Optics Algorithms

Abstract

Rear view cameras are widely used in the automotive industry. They are used in modern car navigation systems to improve the driver’s perception of the situation behind the car. Ultra-wide-angle fisheye lenses are installed on the car for maximum coverage. But such images are not comfortable for human perception. Therefore, one of the main problems in using such cameras is fast algorithms for converting fisheye images into a set of images corresponding to wide-angle and normal virtual cameras, as well as constructing a “top view”. This work examines two image transformation algorithms, both of which provide independent transformations along Cartesian coordinate axes. The first method uses ray tracing through a lens with specified parameters to determine transformation coefficients, the second method provides an empirical selection of coefficients using visual determination of the boundaries of the required fragment interactively. The algorithms are implemented in two versions: the first one is for a regular computer (integrated into the CATIA CAD system) and the second one is built directly into the car rear view camera control processor. The option for a regular computer is used for preliminary elaboration of algorithms, as well as for complete debugging and testing of the rear view camera – both to verify the operation of geometric transformation algorithms and to validate the camera light sensitivity