Vertical seismic profiling (VSP) traveltime tomographic inversion is essentially an underdetermined nonlinear problem. The ray coverage of the acquisition geometry is highly uneven, being dense in shallow and near-well regions but sparse in deep and far-offset areas. In addition, conventional isotropic smoothing regularization tends to blur velocity interfaces while suppressing noise, thereby limiting the accuracy of velocity modeling in structurally complex areas. To address these issues, this paper proposes a first-arrival traveltime tomographic inversion method constrained by stratigraphic structure. In this method, the surface seismic velocity model is used as both the initial model and a background constraint to ensure the stability of the low-wavenumber background of the velocity field during iterations. Stratigraphic boundaries are extracted from seismic migration sections using Canny edge detection, and stratigraphic dips are calculated using the structure tensor enhanced by anisotropic diffusion filtering, thereby obtaining detailed features of structural orientation and interlayer interfaces. Based on the extracted edge and dip information, a stratigraphic-structure regularization term is constructed and incorporated into the tomographic inversion objective function. This enables velocity updates to enhance smoothing along stratigraphic trends while suppressing smoothing across interfaces, thus preserving the sharpness of structural boundaries while maintaining the continuity of the intralayer velocity field. Numerical tests and field data validation from the Kuqa Depression in the Tarim Basin show that the proposed method outperforms conventional isotropic regularization methods in terms of both structural identification clarity and velocity inversion accuracy. Therefore, it can provide effective support for ultra-deep velocity modeling in complex structural areas.