基于图像识别技术白天能见度的算法分析

Algorithm analysis of daytime visibility based on image recognition technology

针对基于图像识别技术白天能见度的探测方法,在大连机场设计了两套能见度探测识别系统,选用了两部分辨率为800万的工业相机和焦距分别为35和50mm的镜头对大连机场出现大雾天气行连续拍照。利用2020年2月8~14日拍摄的白天照片和建设在大连机场跑道东侧的大气透射仪的探测数据,统计分析照片中特定目标区的图像特征与器测能见度的关系,根据柯西米德定律和一元线性回归分析方法,得到了基于图像识别技术探测白天能见度的通用计算方法,该算法与大气透射仪的数据相比误差低于15%,表明基于图像识别技术进行白天能见度的探测是可行的。

Aiming at the detection method of daytime visibility based on image recognition technology,two visibility detection and recognition systems are designed in Dalian airport,two industrial cameras with a resolution of 8million and a focal length of 35and 50mm were used to take continuous photographs of heavy fog in Dalian airport.The relationship between the image characteristics of a specific target area and the visibility measured by the instrument is analyzed statistically by using the daytime photos taken from February 8to 14,2020,and the sounding data of the atmospheric transmission instrument built on the east side of the runway of Dalian airport,based on the Regression analysis′s law and the KOSCHMIEDER law,ageneral calculation method for detecting daytime visibility using image recognition technology is obtained.The error of the algorithm is less than 15%compared with the data of the atmospheric transmission instrument,the results show that it is feasible to detect daytime visibility based on image recognition technology.