摘要
汽车舱内摄像头DMS和OMS是对车内人员的精神状态、生命特征、肢体表达等进行识别的,为了保证算法识别的成功率,对图像中人脸或肢体的亮度有一定要求。针对近红外成像需要主动光源补光的特点,提出了一种汽车舱内摄像头近红外补光的技术方案,依据红外辐射光源空间通量分布、人体表皮的近红外特征、几何光学的折反射规律、Sensor的光电转换能力等方面的理论,建立计算模型,推导出补光LED的功率与图像中人脸亮度DN值之间的关系公式。在数据模型中通过该公式推算应用5.2W近红外补光LED对人脸进行补光后,图像中所拍摄的人脸的亮度DN值进行模拟计算。同时搭建了舱内摄像头的验证系统,通过对试验车内人员实际补光后拍摄图像数据进行亮度DN值分析,并与理论推算结果进行比对,实验结果与计算结果误差小于±5%,满足算法要求,验证了舱内摄像头近红外补光的技术方案的可行性。
The camera DMS and OMS in the car cabin identify the mental state,vital characteristics,and body expression of the occupants.In order to ensure the success rate of the algorithm recognition,there are certain requirements for the brightness of the face or body in the image.In view of the feature of near infrared imaging requiring active light source,a technical scheme of near infrared light supplement for the camera in the car cabin is proposed.A calculation model is established based on the theories of spatial flux distribution of infrared radiation source,near infrared characteristics of human skin,refractive and reflective law of geometric optics,photoelectric conversion ability of Sensor,etc.The relationship formula between the power of the fill light LED and the face brightness DN value in the image is derived.In the data model,the formula is used to calculate the luminance DN value of the face captured in the image after the 5.2 W near infrared supplementary light LED is used to fill the light.At the same time,a verification system is set up for the camera inside the cabin.Through the analysis of the brightness DN value of the image data taken by the test vehicle after the actual light filling,and the comparison with the theoretical calculation results,the error between the experimental results and the calculated results is less than±5%,which meets the algorithm requirements,and verifies the feasibility of the technical scheme of near infrared light flling for the camera inside the cabin.
作者
丁金延
DING Jin-yan(Fulscience Automotive Electronics Co.,Ltd.,Changchun 130000,China)
出处
《激光与红外》
CAS
CSCD
北大核心
2024年第10期1619-1625,共7页
Laser & Infrared
