Three-dimensional (3D) reconstruction technology is widely applied in fields such as industrial manufacturing and cultural heritage conservation. As industrial measurement accuracy requirements increase, vision only algorithms cannot meet metrology-grade demands, while industrial computed tomography (CT) completely lacks surface color and texture information. To address the balance on geometric accuracy and surface texture in 3D reconstruction, a truncated signed distance function (TSDF) surface texture fusion reconstruction framework guided by high-precision CT surface priors is proposed. First, high-precision global spatial registration of heterogeneous sensors is achieved via Zhang′s calibration method and Lie algebra-based optimization. Subsequently, a bounding volume hierarchy (BVH) accelerated M-ller-Trumbore ray-casting algorithm is employed to calculate pixel-level depths from the high-precision CT mesh, circumventing the uncertainty of visual depth estimation. Finally, a TSDF volumetric field is constructed to fuse multi-view red green blue (RGB) images and depth information, realizing 3D model reconstruction with both precise geometric morphology and authentic surface textures. Experimental results demonstrate that on a texture-rich toy model and a highly reflective metal part datasets, the average distance errors of the reconstruction are 0.035 and 0.024 mm, respectively, representing an approximately 73% reduction compared to mainstream neural radiance field (NeRF) and 3D Gaussian splatting (3DGS) methods. The F1-score exceeds 0.950, while the visual metric structural similarity index measure (SSIM) reaches 0.897, and learned perceptual image patch similarity (LPIPS) is as low as 0.145. The single-view processing time is approximately 0.67 s, and the full-model reconstruction time is shortened to the one-minute level, meeting the real-time requirements of industrial online inspection. This method effectively overcomes the limitations of missing textures in CT reconstruction and insufficient accuracy in optical reconstruction, providing a solution with both metrology-grade geometric accuracy and high-fidelity texture for cultural heritage digitization and precision industrial inspection.