Floorplan-Jigsaw: Jointly Estimating Scene Layout and Aligning Partial Scans

TL;DR

This paper introduces a method for aligning partial 3D scans of indoor scenes by jointly estimating room layouts and transformations using floorplan priors, without relying on feature matching, enabling reconstruction of large or featureless environments.

Contribution

The method uniquely leverages the Manhattan World assumption to jointly estimate scene layout and alignment, improving accuracy without feature descriptors.

Findings

Outperforms existing methods in accuracy and robustness.

Effective on real and synthetic scenes of various sizes.

Enables reconstruction of featureless or large scenes.

Abstract

We present a novel approach to align partial 3D reconstructions which may not have substantial overlap. Using floorplan priors, our method jointly predicts a room layout and estimates the transformations from a set of partial 3D data. Unlike the existing methods relying on feature descriptors to establish correspondences, we exploit the 3D "box" structure of a typical room layout that meets the Manhattan World property. We first estimate a local layout for each partial scan separately and then combine these local layouts to form a globally aligned layout with loop closure. Without the requirement of feature matching, the proposed method enables some novel applications ranging from large or featureless scene reconstruction and modeling from sparse input. We validate our method quantitatively and qualitatively on real and synthetic scenes of various sizes and complexities. The evaluations and comparisons show superior effectiveness and accuracy of our method.