A Preliminary Study on Spatial Sampling for Topographic Data

In this paper, an effective spatial sampling method is preliminary explored to assess the quality of the topographic data. The most novel idea in this study is we think the error as noise in the signal and we hope to direct spatial sampling through realizing this idea. First, we perform a prior partition of the original topographic data on the basis of the different error levels after a rough classification. Second, errors are supposed to obey the normal distribution, so we can use different Gaussian filters to remove the errors in each region. Then error-free topographic data after filtering can be seen as the true data. Finally, on the basis of this result, we can obtain the difference between the original topographic data and the error-free topographic data and then configure an error surface. The extreme points on the error surface are seen as the representative of the sampling sets. Finding the extreme points on the error surface is equivalent to selecting the least amount of sample points; meanwhile, these sample points can reflect the true error of the topographic data. By means of the proposed method, we hope to find the most optimum selection of spatial sample points from feature points, which can depict an approximate distribution of the errors, as well as reflect the entire quality of the topographic map.