In the era of big data, from digital cities to digital earth, the exponential growth of spatial information due to the development of diverse data collection technologies has been a significant concern. Delaunay triangulation has garnered widespread attention and application in geomorphological analysis, topographic simulation, and cartographic synthesis due to its minimal data redundancy and excellent stability. However, as the application fields of Delaunay triangular mesh models continue to expand and application requirements deepen, especially with the urgent need to address real-time large-scale scene rendering and terrain visualization, the efficiency, accuracy, and stability of Delaunay triangulation meshes are increasingly demanded. This paper proposes a parallel optimization algorithm based on the insertion point method, following an analysis of the traditional insertion point method, and demonstrates its effectiveness through a series of experiments.