基于散点先验及BG拟合海雾图像浓度检测  

Inspection of sea image fog concentration based on scatter prior and BG fitting

作　　者：温立民[1,2] 朱朝辉 韩颖 王会峰[1,2] WEN Li-Min;ZHU Chao-Hui;HAN Ying;WANG Hui-Feng(School of Electronic&Control Engineering,Chang’an University,Xi’an 710064,China;Electrotechnical and Electronic Learning Center,Chang’an University,Xi’an 710064,China)

机构地区：[1]长安大学电控学院,西安710064 [2]长安大学电工电子教学中心,西安710064

出　　处：《四川大学学报（自然科学版）》2024年第6期83-94,共12页Journal of Sichuan University(Natural Science Edition)

基　　金：国家自然科学基金(52172324);陕西省科技攻关项目(2015GY052);陕西省交通厅重点项目(20-38T);西安市未央区科技计划(202121)。

摘　　要：针对视场雾浓度等级评定缺乏有效图像检测算法的不足,提出一种基于灰度差-比(gray difference-ratio,GDR)的散点图先验实现雾浓度等级评定方法,引入泊松分布确定最佳采样点,采用B.Gompertz模型函数拟合散点图.首先建立不同条件下标准图像集的散点图,提取S型灰度差-比先验,从图像像质退化模型出发,验证了灰度差-比先验的正确性;其次,建立散点图纵向断面集合点的泊松分布函数,通过计算最大概率确定最佳回代样本点;再次,将样本点带入拟合标准图像集并进行估计参数a、b、k;将样本点代入真实雾图像集估计参数a'、b'、k',并建立标准图像集参数估计值的查找表,通过查表方法确定真实雾图浓度等级.经过同场景不同浓度、不同场景不同浓度图像样本测试,算法结果与PM_(2.5)相关系数达0.91,表明算法测试结果符合浓度变化趋势;经不同浓度有雾和无雾的对比图像测试,表明算法能够作为视场雾浓度等级评定;经对比测试证明本文算法在测试精度达到1.9%.Aiming to address the limitations of image inspection in fog-concentration assessment,a scatter⁃plot prior based on gray difference-ratio(GDR)was proposed to achieve accurate fog concentration.Addi⁃tionally,Poisson distribution was employed for determining the optimal sampling point,and the B.Gompertz model function was utilized for fitting the scatter plot.Firstly,scatter charts were utilized to establish standard image sets under different conditions,and the S-type gray difference-ratio prior were selected.The validity of the gray difference-ratio prior was then verified based on the degradation model of image quality.Secondly,a Poisson distribution function for the assembly point of the longitudinal section of the scatter diagram was es⁃tablished in order to identify the optimal sample point.Thirdly,the sample points was substituted into the im⁃age set to estimate parameter a、b、k of standard sets and parameters a'、b'、k' of the real fog map.The fog concentration level of a real image could be determined by referencing an established lookup table.Simula⁃tions validate that the algorithm's test results exhibit a correlation coefficient of 0.91 with PM_(2.5) by tested across various concentrations in different scenes,and experiment also alignment between the algorithm's out⁃comes and concentration trends.These tests demonstrate the capability of this algorithm to assess fog concen⁃tration grades within a given field of view by contrasting images with varying levels of fog and no fog present.The comparative test confirms that this algorithm achieves a testing accuracy rate of 1.9%.

关 键 词：雾 浓度 图像处理 灰度差-比先验 B.Gompertz函数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]