Semiautomatic road extraction using dense GNSS trajectories and histogram correlation

• Grant number: 18/15142-9
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: November 01, 2018
• End date: March 31, 2019
• Field of knowledge: Physical Sciences and Mathematics - Geosciences - Geodesy
• Principal Investigator: Aluir Porfírio Dal Poz
• Grantee: Daniel José Padovani Ederli
• Host Institution: Faculdade de Ciências e Tecnologia (FCT). Universidade Estadual Paulista (UNESP). Campus de Presidente Prudente. Presidente Prudente , SP, Brazil

Abstract

In this research project a method for the semiautomatic road tracing from dense GNSS trajectories is proposed, based on two basic steps: initialization and tracing. The initialization step follows three basic steps: a sequence of points is linearly interpolated between two seed points previously selected by an operator; histograms with amplitude compatible with the width of the road are established at interpolated points along directions that are orthogonal to the road - the absolute frequency of each class coincides with the number of GNSS trajectories contained in the respective amplitude of the corresponding class; a mean (model) histogram is computed via arithmetic mean among the absolute frequencies of the corresponding classes. In the tracing stage the points of the road centerline are extracted sequentially. Each point of the road centerline is extracted as follows: a local trajectory model is established via linear regression, based on the last extracted points; this model is used to linearly extrapolate the last extracted point; at the extrapolated point a histogram with a width slightly larger than the model histogram is extracted; the position of the extrapolated point is refined via 1D correlation strategy based on the quadratic error correlation function. The local trajectory model is updated and the tracing process continues until the end of the road is reached. The method will be evaluated experimentally based on GNSS trajectories to be obtained with the open-access OpenStreetMap project.