Automatic Road Map Encoding by Fluency Function Approximation of Thin Line Images

Abstract

In this paper, an encoding framework of road map information based on automatic Fluency function approximation of thin line images developed in a recent research is introduced. Road map images having multiple structural elements are stratified into layers of 1bit/pixel images based on color. Each of these layers is processed in stages of adaptive pixel sequence tracing, connectivity restoration, and eventually function approximation. In our experiments, numerical maps of 1/200000 scale obtained from the Geographical Survey Institute of Japan are used. The experimental results confirmed the proposed framework is able to encode the numerical map images in which the visual quality is maintained upon decoding. This framework can be a useful preprocessing step in R&D of high quality GIS applications.