cs.CVNov 10, 2026

Sim(3)-TurboReg: Robust and Efficient Scale-Aware Registration via Tri-Scale Consistency

Authors: Zijun Zhou, Shaocheng Yan, Zhenjun Zhao, Yutong Hu, Ke Yang, Yingdong Gu, Pengcheng Shi, Jiayuan Li

Organizations: Wuhan University · University of Zaragoza

Abstract

Feed-forward 3D foundation models recover geometry at high fidelity, yet inference memory constraints keep them from processing long sequences natively. Current streaming systems work around this limit with sliding-window chunking, which introduces severe monocular scale ambiguity. Resolving this ambiguity requires Sim(3) registration that stays accurate and efficient under outlier contamination. We introduce Sim(3)-TurboReg, which reformulates the compatibility graph construction of TurboReg by replacing absolute distance constraints with a strict Tri-Scale Consistency (TSC) criterion. Our analysis shows that TSC suppresses both mismatching and coordinate-distortion outliers. We integrate the resulting solver into existing streaming systems and obtain high-fidelity, kilometer-scale reconstruction for feed-forward models. On controlled unknown-scale registration at 90% outliers, ours matches RANSAC accuracy with 92.5% lower median latency. In DA3-Streaming on KITTI, ours improves mean ATE RMSE by 22.2% and reduces mean end-to-end time by 16.6%. Code: https://github.com/ivan-zzj/Sim3-TurboReg

Figures & tables

Explore similar work

CardsList
  1. R3R^3: 3D Reconstruction via Relative Regression

    May 26, 2026Congrong Xu, Huachen Gao, Xingyu Chen +3Relative Pose EstimationStreaming 3D Reconstruction

  2. Glob3R: Global Structure-from-Motion with 3D Foundation Models

    Jul 10, 2026Junyuan Deng, Heng Li, Kejie Qiu +7Bundle AdjustmentMulti-View 3D Reconstruction

  3. CLoSeR: Closing the Loop for Long-Context Streaming Reconstruction

    Oct 1, 2026Moyang Li, Zihan Zhu, Wei Zhang +23D Foundation ModelsStreaming 3D Reconstruction