Abstract
Dark channel prior (DCP) has advantages in image enhancement and image haze removal and is explored to detect highway visibility according to the physical relationship between transmittance and extinction coefficient. However, there are three major error sources in calculating transmittance. The first is that sky regions do not satisfy the assumptions of DCP algorithm. So the optimization algorithms combined with region growing and coefficient correction method are proposed. When extracting atmospheric brightness, different values lead to the second error. Therefore, according to different visibility conditions, a multimode classification method is designed. Image blocky effect causes the third error. Then guided image filtering is introduced to obtain accurate transmittance of each pixel of image. Next, according to the definition meteorological optical visual range and the relationship between transmittance and extinction coefficient of Lambert-Beer’s Law, accurate visibility value can be calculated. A comparative experimental system including visibility detector and video camera was set up to verify the accuracy of these optimization algorithms. Finally, a large number of highway section videos were selected to test the validity of DCP method in different models. The results indicate that these detection visibility methods are feasible and reliable for the smooth operation of highways.
Highlights
According to the road traffic accidents annual reports of China from 2007 to 2014, more than 50 percent of accidents and deaths are caused by low visibility which is below 200 meters [1]
According to assumptions that there are some dark pixels in local area of image, Dark channel prior (DCP) calculates the transmittance
Images captured by highway video surveillance systems are colored; the first step is to convert these images to dark channel images, which will be segmented into sky region and nonsky region based on region growing method
Summary
According to the road traffic accidents annual reports of China from 2007 to 2014, more than 50 percent of accidents and deaths are caused by low visibility which is below 200 meters [1]. Current video visibility detection algorithms contain four main fields: template matching, camera model calibration, dual differential luminance, and DCP method. For camera calibration model based on the contrast method, Steffens [6] photographed black object image and calculated visibility value according to the relative brightness ratio of object and its background. The contrast curve of these points could reflect road luminance variation They found the maximum distinguishable pixels to calculate the maximum visibility distance. According to assumptions that there are some dark pixels in local area of image, DCP calculates the transmittance When this method is applied directly to the highway. An optimized DCP algorithm is presented which combines region growing, correction coefficient and multimode detection method to obtain more accurate transmittance
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
