The increasing trend towards the use of image sensors in transportation is driven both by legislation and consumer demands for higher safety and a better driving experience. Awareness of the environment that surrounds a vehicle can make driving and manoeuvring of the vehicle much safer for all road users. The authors present an image-processing method to detect lane departures using video taken from multiple optical cameras that is specifically designed to be in accordance with proposed automotive lane departure warning standards. This multi-camera system is more robust to errors caused by lane marking occlusions, sensor failure and glare that single camera systems can suffer from. The system uses a novel lane marking segmentation algorithm in accordance with international standards for lane markings. This method does not demand the high computational requirements of inverse perspective mapping (IPM) unlike methods proposed in related research. The authors present a method for lane boundary modelling based on subtractive clustering and Kalman filtering, which is within the constraints of automotive standards. Finally, using the cameras intrinsic and extrinsic parameters, the width of the vehicle and guidelines issued by the International Organisation for Standardisation, the authors show how lane departure can be identified. Results are presented that verify the system's high detection rate and robustness to various background interference, lighting conditions and road environments.