A shape analysis and template matching of building features by the Fourier transform method

Shape cognition and representation play an important role in spatial analysis because shape contains some characteristics of geographic phenomena that can be mined to discover hidden geographic principles. As a difficult cognition problem, the shape representation problem in GIS field has the properties of abstraction, indetermination and symbolization. How to use a model to represent shape cognition in our mental world and how to use a single number to compute the shape measure are interesting questions. In the image processing domain, there are many shape measure methods, but there are few proposals for corresponding vector data. This study aims to build a polygon shape measure and offers a Fourier transform-based method to compute the degree of shape similarity. The procedure first represents the boundary of the vector polygon shape as a periodic function, which is expanded in a Fourier descriptor series, and then, it obtains a set of coefficients that capture the shape information. Through the experiment on spatial shape match and shape query, the study shows that Fourier transform-based shape identification and template matching is consistent with human cognition.