A point-feature label placement algorithm based on spatial data mining.

Area of Science:

• Geographic Information Science
• Computational Geometry
• Data Mining

Background:

• Automated point-feature label placement (PFLP) is crucial for clear map generation.
• Existing PFLP methods often struggle with dense datasets due to insufficient consideration of spatial distribution and label correlations.
• This leads to suboptimal label quality in complex point datasets.

Purpose of the Study:

• To propose an improved PFLP algorithm leveraging spatial data mining techniques.
• To address limitations in handling spatial distribution characteristics and label correlations in dense point datasets.
• To enhance the quality and efficiency of automated map label placement.

Main Methods:

• Developed a point-feature label placement algorithm based on spatial data mining.
• Introduced a label frequent pattern framework (LFPF) to quantify feature interference.
• Employed metaheuristic algorithms (simulated annealing, genetic, ant colony) with the LFPF for validation.
• Proposed a bit-based grid spatial index to optimize conflict detection.

Main Results:

• The proposed method significantly improved label quality compared to existing algorithms and recent literature.
• Label quality enhancements ranged from 3 to 6.7 and 0.1 to 2.6 respectively.
• Label placement efficiency increased by 58.2% compared to traditional grid indexes, with reduced memory and time consumption.

Conclusions:

• The spatial data mining approach effectively addresses challenges in dense point-feature label placement.
• The LFPF and bit-based grid index contribute to superior label quality and efficiency.
• This algorithm provides a robust solution for automated, high-quality map label generation.