隧道衬砌图像清晰度影响因素的模型试验研究

Model test study on factors affecting image sharpness of tunnel lining

针对隧道衬砌图像清晰度,通过移动机械平台搭载线阵相机、光源(LED或LASER)、工业电脑等设备开展室内模型正交试验,采用一种融合图像幅度谱斜率和图像总变差的图像清晰度量化算法,以方差分析的手段研究光源类型、物面照度、曝光时间、物面直径等四个影响因素的显著性及优水平,并分析不同水平下图像清晰度的变化规律。试验结果表明:影响图像清晰度的主次因素顺序是曝光时间、光源类型、物面照度、物面直径,其中物面直径为不显著因素;试验因素水平下最优组合为LED,6 klux,90μs。研究成果可为基于计算机视觉和机器学习的隧道病害检测装备的研制提供参考依据。

Using orthogonal experiment analysis,this paper presents a model test study on factors affecting image sharpness of tunnel lining via a mobile platform equipped with linear array camera,light of either LED or LASER,industry computer etc.Four factors are considered in this paper,which are type of light source,illumination intensity of object surface,exposure time and surface diameter.An algorithm which utilizes both the slope of the magnitude spectrum and the total spatial variation is implemented to measure the local perceived sharpness in some images acquired in the test.Significance and optimal levels of the four factors are calculated by analysis of variance.We also discuss about changing laws of image sharpness under different testing factor levels successively.The test results show that the order of principal factors affecting image sharpness are exposure time,type of light source,illumination intensity of object surface,surface diameter,among which surface diameter is an undistinguished factor.The best combination of optimal levels is LED,6 klux,90 μs.This study can show a delighted help for inspection equipment of tunnel lining defects based on CV and ML.