Precision and Bias of Data from Film

Abstract

An objective of the Long-Term Pavement Performance study is the development of improved pavement design procedures. This objective depends on performance data being collected on present in-service pavements. The primary pavement distress data collection was done by recording 35mm film images at intervals since 1988. Five rounds of images have been collected. The accuracy and repeatability of distress data collection was of concern. The effort to assess the quality of distress data resulting from analysis of 35mm pavement images is reported. A set of 12 pavement sections, including asphalt surface, jointed and continuously reinforced concrete, was digitized three times by 10 operators with various skill levels at intervals of several weeks between passes. The distress analysis was done by using four semiautomated computerized workstations to view a projected image of the pavement on a digitizing tablet with the operator using a cursor to input distress type, severity, and location. A data summary and plot were output from the software. The results of each digitizing pass were used to evaluate the repeatability of the operators, and grouped data were used to evaluate accuracy.