Route Matching in Sketch and Metric Maps

Despite all distortions and incompleteness present in Sketch maps, they can be used for different purposes such as collecting and searching spatial data and organizing the spatial knowledge of individuals. Since the prerequisite for all of these cases is matching sketches with existing data in spatial databases or metric maps, several studies have been conducted in this field. These studies generally focus on the relationships between routes, landmarks, and regions, while data sources do not necessarily contain all of these data. Also, the subject of the incompleteness of data and evaluation of the results has been less considered in previous studies. Therefore, in this paper, a new solution for matching routes from sketches to metric maps is presented. The proposed solution uses the Genetic Algorithm to measure the similarities between routes in two phases, and in each phase, a set of geometric/topological criteria is considered to improve the matching. The proposed algorithm is implemented to match routes of 25 sketches with the routes in a metric map. The average FMeasure\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$F_{{{\text{Measure}}}}$$\end{document} of matching is 45.59%. The results show that by considering the specific conditions (not availability of descriptive data of routes and geometric/descriptive information of landmarks), this paper has been able to take an important step toward finding an acceptable solution for the matching problem. The proposed methods can be used to deal with the incompleteness of the thematic information in metric maps.