Leveraging Inlier Correspondences Proportion for Point Cloud Registration
This work addresses the problem of accurate point cloud registration for applications like robotics and 3D reconstruction, offering an incremental improvement over existing methods.
The paper tackles the challenge of constructing correct correspondences in feature-learning based point cloud registration, especially with partial inputs and indistinguishable surfaces, by proposing a Geometry-guided Consistent Network (GCNet) that leverages inlier correspondences proportion through techniques like a pyramid hierarchy decoder, consistent voting strategy, and geometry guided encoding module. Experiments on indoor, outdoor, and synthetic datasets show that GCNet outperforms state-of-the-art methods, with the techniques being model-agnostic and easily migratable to other methods.
In feature-learning based point cloud registration, the correct correspondence construction is vital for the subsequent transformation estimation. However, it is still a challenge to extract discriminative features from point cloud, especially when the input is partial and composed by indistinguishable surfaces (planes, smooth surfaces, etc.). As a result, the proportion of inlier correspondences that precisely match points between two unaligned point clouds is beyond satisfaction. Motivated by this, we devise several techniques to promote feature-learning based point cloud registration performance by leveraging inlier correspondences proportion: a pyramid hierarchy decoder to characterize point features in multiple scales, a consistent voting strategy to maintain consistent correspondences and a geometry guided encoding module to take geometric characteristics into consideration. Based on the above techniques, We build our Geometry-guided Consistent Network (GCNet), and challenge GCNet by indoor, outdoor and object-centric synthetic datasets. Comprehensive experiments demonstrate that GCNet outperforms the state-of-the-art methods and the techniques used in GCNet is model-agnostic, which could be easily migrated to other feature-based deep learning or traditional registration methods, and dramatically improve the performance. The code is available at https://github.com/zhulf0804/NgeNet.