Abstract
The distribution of multiclass discrete data in geographic space is a research hotspot in the field of geographic-related visualization. As a basic visual presentation of such data, the advantages of dot maps are perceptual intuition and abundant details, but there is also the problem of poor readability due to the points overlap. The approach of density estimation by resolution is proposed in this paper to optimize dot maps, and to flexibly adjust sampling parameters of the current resolution, so as to show the details to the maximum extent and maintain the relative density characteristics of various types of property. In order to compensate the missing discrete features caused by sampling, a series of interactive tools are used to effectively improve the accuracy of visual analysis and assist the overall visual representation. Finally, the effectiveness of this approach is proved through case analysis and user research.