关键词: 点要素注记, 聚类分组, 蚁群算法, 注记配置方案, 注记适应度评价函数

Abstract:

The problem of Large-scaled Point Feature Cartographic Label Placement (LS-PFCLP) is one of map labeling difficulties, which is mainly limited by time efficiency and labels’ quality. Changing this plight will accelerate the application development of Intelligent Cartography. The article firstly contraposes the form simplification of traditional potential label position scheme and puts forward an oval multi-orientations and multi-levels cartographic potential label position scheme to make a better conformation of the potential label position scheme to the actual demand, and make it parameterized and multiplex. Secondly, the article combines the space distribution features of point features to adopt an Ant Colony Algorithm based on cluster grouping (C-ACA), whose core framework is decreasing the large scale of points to plural mini-scaled point collections through DBSCAN clustering, then integrating Ant Colony Algorithm to C-ACA. In the process of C-ACA’s implementation, we discuss and optimize the core parameters, such as Eps and MinPts of DBSCAN, ant colony size parameters (\begin{array}{c}{P}_a\end{array}、\begin{array}{c}{\eta }_{\mathit{cj}}\end{array}、\begin{array}{c}{\tau }_{\mathit{cj}}\end{array}、\begin{array}{c}{P}_{\mathit{cj}}^b\end{array}), and evaluation function of \begin{array}{c}{E}_{\left(\mathit{bc}\right)}\end{array}, to achieve Large-scaled Point Feature Cartographic Label Placement. Experiments show that C-ACA has a great contribution to the efficiency improvement of LS-PFCLP. When compared, with two cases of random points tests and one realistic boundary points case, with techniques such as Ant Colony Algorithm (ACA), the C-ACA has been proven to be an efficient choice, with better performance in both efficiency and quality. In case 1, The C-ACA improves the efficiency by 73.2 percent than normal-ACA, when the label density falls in between 5% and 30%. Moreover, the altered resultant label’ quality is 8.0 percent better than before. In case 2, the result of C-ACA has justified the stability presentation. In case 3, we have used this algorithm to process the boundary points of the collective land ownership data of Fushun county in China, which has improved the efficiency by 86.7% and the quality by 14.6% with outstanding performance. We concluded that the improved algorithm is applicable to LS-PFCLP with points that having massive quantity and greater variations of clustering density.

Key words: the point feature cartographic label, cluster grouping, ant colony algorithm, label cartographic pattern, the label fitness evaluation function