An MPI-based parallel genetic algorithm for multiple geographical feature label placement based on the hybrid of fixed-sliding models

TL;DR

This paper introduces an MPI-based parallel genetic algorithm that efficiently solves the NP-hard problem of geographical feature label placement, significantly reducing computation time and improving label quality in geographic information visualization.

Contribution

The study presents a novel MPI parallel genetic algorithm for MGFLP based on a hybrid fixed-sliding model, addressing computational complexity and enhancing label placement efficiency.

Findings

Reduces computational time from over 20 minutes to less than 1 minute.

Achieves fewer label-feature conflicts compared to previous methods.

Improves overall quality score of label placement.

Abstract

Multiple geographical feature label placement (MGFLP) has been a fundamental problem in geographic information visualization for decades. The nature of label positioning is proven an NP-hard problem, where the complexity of such a problem is directly influenced by the volume of input datasets. Advances in computer technology and robust approaches have addressed the problem of labeling. However, what is less considered in recent studies is the computational complexity of MGFLP, which significantly decreases the adoptability of those recently introduced approaches. In this study, an MPI parallel genetic algorithm is proposed for MGFLP based on a hybrid of fixed position model and sliding model to label fixed-types of geographical features. To evaluate the quality of label placement, a quality function is defined based on four quality metrics, label-feature conflict, label-label conflict, label ambiguity factor, and label position priority for points and polygons. Experimental results reveal that the proposed algorithm significantly reduced the overall score of the quality function and the computational time of label placement compared to the previous studies. The algorithm achieves a result in less than one minute with 6 label-feature conflicts, while Parallel-MS (Lessani et al., 2021) obtains the result in more than 20 minutes with 12 label-feature conflicts for the same dataset.