A Constrained Optimization Approach for Gaussian Splatting from Coarsely-posed Images and Noisy Lidar Point Clouds

Jizong Peng

Abstract:

D Gaussian Splatting (3DGS) is a powerful reconstruction technique; however, it requires initialization from accurate camera poses and high-fidelity point clouds. Typically,
the initialization is taken from Structure-from-Motion (SfM)
algorithms; however, SfM is time-consuming and restricts
the application of 3DGS in real-world scenarios and largescale scene reconstruction. We introduce a constrained optimization method for simultaneous camera pose estimation
and 3D reconstruction that does not require SfM support.
Core to our approach is decomposing a camera pose into a
sequence of camera-to-(device-)center and (device-)centerto-world optimizations. To facilitate, we propose two optimization constraints conditioned on the sensitivity of each
parameter group and restricts the search space of each parameter. In addition, as we learn the scene geometry directly from the noisy point clouds, we propose geometric
constraints to improve the reconstruction quality. Experiments demonstrate that the proposed method significantly
outperforms the existing (multi-modal) 3DGS baseline and
methods supplemented by COLMAP on both our collected
dataset and two public benchmarks. Project webpage:https://eldentse.github.io/contrainedoptimization-3dgs.