Abstract: DEM is the foundation in visualization of 3D terrains. With the sharp increase of DEM databases, the compression of DEM has become an important research on the visualization of 3D terrains. Based on the entropy coding, arithmetic can preserve the significant detail information. This paper studies on arithmetic coding and divides those predictive models into three types: the simple linear predictive model, the Lagrange predictive model and the least square predictive model. This paper draws a comparison among them. This paper points out some problems when arithmetic coding is used in practice and makes an expectation for the future research.

Key words: Digital Elevation Model（DEM）, arithmetic coding, lossless compression, predictive model

摘要： DEM是三维地形可视化基础，随着DEM数据量的不断增加，对DEM进行编码压缩已成为三维地形可视化的重要研究内容。算术编码是一种基于熵编码的无损压缩编码，能保留重要细节信息。目前基于算术编码的预测模型可分为简单线性预测、拉格朗日预测和最小二乘预测三类，对这三类算法进行了对比分析。指出了算术编码算法在实际运行中存在的问题，对其未来发展提出展望。

关键词: 数字高程模型（DEM）, 算术编码, 无损压缩, 预测模型