Roads system is represented as graph containing vertices (intersections) and edges (road segments). Vertices are represented as id with associated longitude, latitude (arc sec accuracy). Edges are polygonal chain with ID, address range and an array of vertices.
We get data from several sources and would like to merge graphs in one.
Picture below shows typical situations.
Distances between lines, curvature, angle are useful parameters to estimate similarity. $\int_{a}^{b} (f(x)-g(x))^2 dx$ would be useful, but how can I project line segments of different length on each other?
What would be agood metrics to detect similar line segments?
Found answers to similar questions. There are many options:
https://en.wikipedia.org/wiki/Fr%C3%A9chet_distance
https://en.wikipedia.org/wiki/Dynamic_time_warping
https://en.wikipedia.org/wiki/Procrustes_analysis
Discovering Similar Multidimensional Trajectories Curve Matching, Time Warping, and Light Fields: New Algorithms for Computing Similarity between Curves
APPLIED SIMILARITY PROBLEMS USING FRECHET DISTANCE