LANE INDICATION/RECOGNITION SYSTEM BY USING IMAGE SENSORS: EVALUATION OF INDICATION PERFORMANCE OF LANE MARKS IN THE REAL WORLD

Studies of motor vehicle automatic driving technology, using image sensors such as CCD cameras, are underway. In order to establish automatic driving technology, which is based on image sensors, improving lane recognition performance under a wide variety of meteorological environments is a key technology. To improve the reliability of image sensors, we decided to investigate the following three areas: (1) improving the performance of the CCD image sensor, (2) developing recognition logic, (3) evaluating the visibility(indication performance) of lane marks. In order to quantify lane recognition performance, we designed a measurement system for the contrast in luminance between lane marks and the road surface. Experimenting with this measurement system in the real world was an effective way of evaluating lane recognition performance. The contrast in luminance of the lane marks changed greatly according to the type of road surface and the meteorological conditions. We therefore propose a performance requirement for the image sensor to improve the reliability of the lane recognition performance and the improvement item of a traffic environment.