Experiment of Multispectral Sensing Sensor for Urban Road Materials in Outdoor Environment

Abstract

This research is the first attempt to conduct several experiments of multispectralsensing sensor for urban road materials in outdoor environment. This research aims to classifyfive urban road materials that are aggregates, asphalts, concrete, clay, natural fibre includingvegetation and water. There were 9 cameras in the multispectral sensing sensor. Seven cameraattached with narrow band optical filter with the centre spectrum at 710nm, 730nm, 750nm,800nm, 870nm, 905nm and 950nm. One camera attached with 720 nm normalization band useshigh pass optical filter. Another camera attached with UV/IR cut optical filter works as a RGBcamera. The images results, that have been taken, are processed in MATLAB to get the imagingindex results from the multispectral system. Naïve Bayes classifier is used in Weka to classifythe urban road materials with vegetation and water. The first classification and testing thatclassifies five urban road materials with vegetation and water have accuracy results ranged from0 % to 32% while the accuracy results without vegetation and water have better accuracy resultsranged from 0 % to 55 %.