Abstract
This paper addresses the effective and quantitative image DB construction for the development of front looking camera systems. The automotive industry has expanded the capability of front camera solutions that will help ADAS(Advanced Driver Assistance System) applications targeting Euro NCAP function requirements. These safety functions include AEB(Autonomous Emergency Braking), TSR(Traffic Signal Recognition), LDW(Lane Departure Warning) and FCW(Forward Collision Warning). In order to guarantee real road safety performance, the driving image DB logged under various real road conditions should be used to train core object classifiers and verify the function performance of the camera system. However, the driving image DB would entail an invalid and time consuming task without proper guidelines. The standard working procedures and design factors required for each step to build an effective image DB for reliable automotive front looking camera systems are proposed.