| Summary: | Background. The growth of mobile technologies led to the ability to survey the
condition of road pavement on streets and highways using data received from sensors
of mobile devices, which are not always reliable information sources. Thus, this
work is focused on the improvement of road pavement defect detection in case of
missing fragments in the initial data.
Materials and methods. The research proposes an approach to pavement defects
detection based on the analysis of data collected by mobile devices. Due to the complexity
of processing and transmission of video recordings from a mobile device's
camera, the proposed approach uses data from an accelerometer and a
GPS/GLONASS receiver. The unreliability of the accelerometers and possible failures
in the transmission of measurement results over the cellular network lead to
missing fragments in the initial data, that is why the proposed approach is based on
the sequential implementation of Wavelet transform and intelligent neural network
analysis, which together provides an increase in the probability of pavement defects
detection.
Results. Based on the proposed approach, a system for survey the state of the
road pavement was developed. Experimental testing of the system was carried out
on two sections of the road network. Test sections are different in the quality of the
pavement and speed modes.
Conclusions. The results of the study showed that the probability of road pavement
detection was 0.95, thus, high rates were achieved relative to similar solutions.
|
|---|
