Measures for Matching Stereo Images in Road Pavement Diagnostics

The paper addresses the overall body of problems related to application of measures for matching stereo images in the measuring instruments used for assessment of road infrastructure condition. Selected stereo vision solutions applied in road pavement diagnostics have been explained. The main problem connected with the subject in question, namely the ambiguity of image matching, has been briefly described, and the matching algorithms commonly used have been characterised. The following measures of matching have been discussed: Sum of Absolute Differences (SAD), Zero Mean Sum of Squared Differences (ZSSD) as well as Covariance-Variance (CoVar). The studies conducted in the field addressed in the article concerned efficiency of stereo image matching for the aforementioned measures of matching. The relevant measurements were performed using images developed as outcomes of field surveys of the road infrastructure condition. The three-dimensional road pavement mapping thus obtained was compared with the mapping based on direct measurements using a dedicated diagnostic station.