Research On Nighttime Vehicle Light Identification Applied To Adaptive Driving Beam System

Automotive lighting system plays an important role in the safety of driving at night.As one of the important components of the lighting system,the automotive high beam improves the line of sight and it expands the field of view.It has an irreplaceable function especially in the case of very poor lighting conditions.However,the abuse of high beam lights when meeting and following cars can also cause dazzling and it affects the sight line of others,which seriously affects traffic safety.With the increasing use of LED lamps in automobiles,the adaptive driving beam(ADB)system for LED headlamps is also developing rapidly to solve the problem of glare caused by high beam lights.The adaptive driving beam system eliminates the glare effects caused by the high beam to the vehicle in front while obtaining an adequate viewing field.The most important thing in this system is how to quickly and accurately identify the vehicle in front at night.The main contents of this paper are as follows:1)The environmental conditions are analyzed while the adaptive driving beam system is working.The imaging characteristics of vehicles at night are also analyzed to determine the technical route of night light recognition.The height range of front car lights in the pixel coordinate system was derived by combining camera imaging principles and national standards.Then the area was divided where the headlights may be present.2)Various digital image noise's mathematical models and their corresponding noise reduction methods were analyzed.The corresponding noise reduction algorithm was designed according to the image collected in the actual experiment and the noise reduction effect was verified through the experiment.3)The advantages and disadvantages of multiple clustering algorithms were analyzed.According to the characteristics of car lights in the picture,the light sources were identified by the grid clustering algorithm.The grid clustering algorithm flow was designed and it was verified by experiment.4)The taillights were identified by halo color according to taillight imaging features.A mathematical model of color calculation was proposed and segmentation threshold was determined based on pixel point statistical features.5)The appropriate image sensor and processor were selected according to the system requirements.Experiments were designed to verify algorithm feasibility.The function of the adaptive driving beam system was presented on a car.The experimental results show that the method proposed in this paper can accurately identify the vehicles' headlights and taillights and it can enable real-time availability in embedded systems.The method can meet the practical requirements and it has important practical significance.It also provides a technical platform for future in-depth research and optimization.