A digital elevation model (DEM) clustering simplification algorithm based on curvature entropy and the Gaussian mixture model

Since the traditional simplification algorithm of the digital elevation model (DEM) makes it difficult to identify the characteristics of complex topography, a new clustering simplification algorithm based on the Gaussian mixture model and curvature entropy is proposed. Firstly, the DEM data are clustered by the elevation information. The combination of curvature and entropy is taken as the simplification feature allowing effective geometric information extraction. For different subclasses of DEM, different DEM simplifications are executed according to the curvature entropy. The GMM algorithm employed for clustering ensures all kinds of terrain data can be retained after simplification, and the flat area is not simplified to an extent which would result in a discontinuity. The experimental results show that compared with traditional simplification algorithms, the algorithm proposed has the characteristics of higher precision, smaller voids and higher terrain retention, making it more suitable for the complex and diversified DEM.