The Compensation Method Of Tunnel Brightness Based On Spatial-temporally Visual Content-preserving

As a special part of the whole highway network,highway tunnel has a high accident rate.Based on investigation,driving behavior of drivers and the frequent tunnel accidents have a direct connection with tunnel transport system,and drivers’ driving behavior is directly affected by their visual comfort level when driving.Aiming at the requirements of a high comfort level of driving,the drivers’ visual characteristics and the visual adaptation process when driving through a tunnel have important practical significance for tunnel safety.Thus,a model of visual brightness compensation is established to improve the visual comfortability when driving.The model is different from previous methods only using spatial compensation and can provide better sequentially visual brightness information.On this basis,it can effectively realize the visual smooth transition at the entrance and exit of highway tunnels,meanwhile provides a new insight for tunnel lighting and transition zone setting,so as to lay the foundation for tunnel safety.The main research contents are as follows:Firstly,this proposal analyses the characteristics of the tunnel traffic environment,accidents and drivers’ visual perception.In the meanwhile,the relationship among the tunnel objective environment,drivers’ visual comfort level and tunnel driving safety is exploited.Aiming at the tunnel visual scene data,this proposal proposes two methods for drivers’ visual comfort level characterization in view of visual sequential brightness change,Just Noticeable Difference Number and SOE.Secondly,a model of tunnel brightness compensation based on Spatial-Temporally Visual Content-Preserving(STVCP)is established,and the compensation process of this model includes three parts: feature extraction and matching,the alignment of video frames and the fusion of video frames.This proposal mainly focuses on the alignment method from the aspects of alignment accuracy and shape preservation.By exploiting the features on some visual brightness compensation methods,including Global Homography,As-Projective-As-Possible,Shape-Preserving Half-Projective,APAP+SPHP is modeled for a better compensation.Specially,in one video,it needs to provide a reference frame for brightness compensation and uncomfortable frames excepting needing compensation.Specifically,SOE and JndN are utilized to determine the reference frame for brightness compensation and the uncomfortable frames,respectively.Last,the reference frame and the uncomfortable frames are fed into the compensation model.In this proposal,extensive visual brightness compensation experiments on visual scenes of tunnel have been conducted.The performance comparison of seven compensation methods is evaluated from the aspect of visual quality,objective quality of video frame and comfortability of sequential vision brightness.Experimental results show that APAP+SPHP modeled by ourselves can obtain more stable and comfortable visual sequential brightness than the others,and has important application value for tackling the visual perception of the “black and white” effect in tunnels.