Analýza provozu na pozemních komunikacích pomocí automatizovaného hodnocení videozáznamu

Abstract

In terms of its objective and propositions, this doctoral thesis focuses on establishing a methodology for evaluating the basic behavior of traffic stream and conflict situations using an automated method of video-analysis. The introductory section offers a review of the current status of knowledge based on the study of technical literature published in the Czech Republic and worldwide. It contains an analysis of knowledge concerning traffic stream behavior and utilization of a traffic conflict technique for road traffic safety evaluation. It informs about the methods of measuring both the frequency and severity of traffic conflict situations using surrogate safety indicators, their predictive value and suitability for various kinds of traffic situations. It also summarizes the new and non-traditional approaches to road safety research using advanced methods enabled by the present development of science and technology. Another part of the research is dedicated to the present status of automated video-analysis development utilizing automated road user detection, and evaluation of their mutual interactions. Based on the acquired information evaluation, further research of the presented thesis is focused on utilization of a semi-automated method of video-analysis using the SAVA software program, borrowed for these purposes thanks to the understanding of the Royal Institute of Technology, Stockholm. The main part of the thesis deals with application of the semi-automated method of video-analysis for measurement at two selected intersections. It uses a computer-aided calculation for both the traffic stream behavior parameters as well as the PET (Post-Encroachment Time) and TTC (Time-to-Collision) safety indicator values. In the conclusion it compares the measurement results using the semi-automated video- analysis with the results of evaluation of the same video-recorded traffic situations using a different method of conflict situation analysis by a trained observer. Based on the comparison of the results measured by both procedures, it presents a proposal for utilization of the results for the further development in this area of science, together with a proposal for possible improvements of the methodology employed and for its practical application.