In traditional multi-line laser 3D reconstruction technology, due to the inevitable noise affecting the multi-line laser lines, the extracted laser center coordinates often contain specific errors. These errors can lead to the inability to obtain high-precision 3D data when using matching points found based on epipolar constraints for 3D reconstruction directly. To address this issue, we propose a method based on geometric estimation to achieve 3D reconstruction of multi-line lasers. First, by calibrating the quadratic surface equations of the multi-line laser, combined with the binocular epipolar constraint method, the initial matching points of the multi-line laser can be calculated. After finding the correct initial matching points, a geometric distance minimization estimation model can be established using the distance constraint from points to epipolar lines. This geometric distance refers to the distance from the laser center points in the left and right images to their corresponding epipolar lines. New matching points that better conform to the epipolar constraints can be recalculated through this geometric distance minimization optimization estimation. Finally, these new matching points can be used to complete the 3D reconstruction of the multi-line laser. Compared to the traditional method based on epipolar constraints, the algorithm proposed in this paper performs better in matching and accuracy. The accuracy of the final 3D reconstruction can reach about 0.02 mm. This method can significantly improve the overall accuracy of binocular multi-line laser reconstruction, thereby obtaining more accurate and reliable 3D data.