An Automatic Visibility Measurement System Based on Video Cameras

This report represents the results of research conducted by the author and does not necessarily represent the views or policy of the Minnesota Department of Transportation. This report does not constitute a standard, specification, or regulation. ***The original form of this report exists only as a printed copy. This electronic version was reproduced for electronic distribution by Dr. Taek Kwon from the original Mn/DOT's printed copy. Readers should be aware that the page numbers do not exactly match with the printed version. While conversion, minor editing was also done but without changing the original contents. This report describes a two year study on visibility measurement methods using video cameras. A good theoretical basis and practical methods along with the experimental results are presented. Among several methods and algorithms developed, the edge decay model along with a proper threshold technique worked best for evaluating daytime visibility. This approach estimates the distance where an object of specified size and shape is no longer distinguishable from the background in terms of edge information. For night time, a constant light source is required to evaluate visibility. A light diffusion model that follows an exponential decay curve was developed. It was found that the volume of light diffused out of the original source is logarithmically correlated to visibility. The developed day and night algorithms were implemented in the field and evaluated using manual measurements. For daytime, visibilities measured using the edge decay model closely approximated the manual measurements on all types of weather. For night time, evaluation was very difficult because of unreliability of manual measurements at night. However, it was verified that the trend of visibility change obtained by the proposed approach closely approximates the trend of manual measurements. ACKNOWLEDGEMENTS The author expresses appreciation to the Office of Research Administration and Mn/DOT District-1 for financial and manpower support for this research. Special thanks to Mr. Edward Fleege of Mn/DOT for his tireless help on all phases of the project. Several maintenance engineers at the Mn/DOT Duluth Office also provided valuable assistance. Thanks also to Mr. Tim Garret of Castle Rock Consulting Company for providing me large collection of video data from the previous visibility study.