Handling of Characteristics of Image Sensors for Driver Assistance systems

Automotive computer vision recognizes objects, can determine properties, classifies objects, and makes decisions based on this information. This talk concentrates on the image sensor in ADAS cameras, the image sensor’s key performance indicators and the part-to-part variation of these figures. Based on established standards, a method is proposed which follows the rules of static and dynamic part average testing. The proposed method allows to predict the image quality for mass production quantities or correlated fluctuations, so that image quality can be maintained, controlled and simulated before the release of a computer vision system and ensured during production of an image sensor.