以格网DEM数据为基础,生成具有良好保凸性、光滑性、平顺性特征的高质量等高线,已逐渐成为制作等高线图的重点。鉴于现有技术方法在沟道、山脊等地形变化部位生成的等高线中存在生硬、相交、异常闭合等问题,本文以黄土丘陵1:5万不同格网分辨率DEM为例,采用对比分析方法,建立了基于DEM格网加密技术的高质量等高线生成方法。实验表明:(1)采用以4×4DEM格网单元为搜索圆的完全规则样条函数插值法,以可视化时图面0.1mm的实际距离为加密DEM的格网分辨率值,对DEM格网进行加密处理后生成的等高线具有更为理想的保凸性、光滑性和平顺性特征;(2)对于已知综合尺度的地形,存在最优DEM格网分辨率阈值,若要高质量地再现原始等高线,则实际DEM格网分辨率不得低于最优DEM格网分辨率阈值。
DEM-based generation of high quality contours with properties of convexity, smoothness and flatness has gradually become a key research area for creating contour maps. The existing generation methods have some deficiencies including hardness, intersection and abnormal closure of contours generated in the terrain change areas, such as channels and ridges. In view of the above-mentioned facts, taking a 1:50 000 DEM in the loess hilly region as the test data set, this paper analyzes the basic conditions of reconstruction contours based on densified DEM grid with high quality and efficiency, and examines the accuracy of different interpolation approaches in the process of this reconstruction. Utilizing the comparative analysis, this paper proposes a novel method to generate and represent contours with high quality based on densified DEM grid scheme and terrain shading technique. Experiment results show that: (1) the interpolation approaches include bilinear function, cubic convolution, local quadratic polynomial, and completely regularized spline function. In view of the extreme values, the discrete degree, and the spatial distribution of elevation numerical error, regularized spline function is the optimal interpolation method to reconstruct surface morphology, followed by local quadratic polynomial, cubic convolution, and bilinear interpolation. (2) The contours, which are generated by densified DEM grid via the complete regularized spline function with taking 4 - 4 DEM grid cells as a search circle, are more accurate and smoother. In the above process, the resolution of densified DEM grid is the actual distance of 0. lmm in the visualized maps. (3) For the terrain with a known generalization scale, there exists an optimal resolution threshold for the DEM. To reproduce the original contours with high quality, the corresponding resolution of DEM cannot be lower than the optimal threshold. (4) The superposition technique between contours and DEM hillshadings attached with a white mask could effect