With the wide application of mesh surfaces in major and complex projects, high-quality mesh subdivision and optimization have become the key to ensuring the accuracy of surface simulation. This paper proposes a mesh surface smoothing optimization method based on Kriging interpolation and adaptive midpoint subdivision to enhance the uniformity, smoothness and aesthetics of the mesh while strictly maintaining the original geometric features. This method first adopts the improved Bowyer-Watson algorithm to dynamically screen the longest edge based on the degree of mesh subdivision or the mesh length limit constraint, gradually insert the midpoint and ensure that the minimum cavity contains the original point, achieving the initial mesh subdivision. Secondly, the positions of the newly added mesh nodes are adjusted through Kriging interpolation technology to enhance the uniformity and adaptability of the mesh. Experiments show that compared with traditional methods, this method has significant improvements in mesh uniformity and boundary preservation, and is particularly suitable for high-precision optimization of complex engineering surfaces. This algorithm has been implemented in the GeoStation geological 3D survey and design system, featuring good engineering applicability and operability, and providing a reliable technical solution for the efficient optimization of complex mesh surfaces.