Abstract:This article introduces three-filters-to-normal $+$ (3F2N $+$ ), an extension of our previous work three-filters-to-normal (3F2N), with a specific focus on incorporating discontinuity discrimination capability into surface normal estimators (SNEs). 3F2N $+$ achieves this capability by utilizing a novel discontinuity discrimination module (DDM), which combines depth curvature minimization and correlation coefficient maximization through conditional random fields (CRFs). To evaluate the robustness of SNEs on noisy data, we create a large-scale synthetic surface normal (SSN) dataset containing 20 scenarios (ten indoor scenarios and ten outdoor scenarios with and without random Gaussian noise added to depth images). Extensive experiments demonstrate that 3F2N $+$ achieves greater performance than all other geometry-based surface normal estimators, with average angular errors of 7.85 $^\circ$ , 8.95 $^\circ$ , 9.25 $^\circ$ , and 11.98 $^\circ$ on the clean-indoor, clean-outdoor, noisy-indoor, and noisy-outdoor datasets, respectively. We conduct three additional experiments to demonstrate the effectiveness of incorporating our proposed 3F2N $+$ into downstream robot perception tasks, including freespace detection, 6D object pose estimation, and point cloud completion. Our source code and datasets are publicly available at https://mias.group/3F2Nplus. Note to Practitioners—The primary motivation behind this work arises from the need to develop a high-performing surface normal estimator for practical robotics and computer vision applications. While geometry-based surface normal estimators have been widely used in these domains, the existing solutions focus merely on discontinuity discrimination. To tackle this problem, this article introduces a plug-and-play module that leverages both depth curvature and correlation coefficient to quantify discontinuity levels, thereby optimizing surface normal estimation, particularly near or on discontinuous regions. Moreover, this article also introduces a large-scale public dataset with random noise added to depth images, providing a more realistic and robust platform for algorithm evaluation within this research community. Extensive experimental results demonstrate that our method outperforms other state-of-the-art algorithms.

Adaptive Surface Normal Constraint for Geometric Estimation From Monocular Images

Adaptive Surface Normal Constraint for Geometric Estimation from Monocular Images

Geometry Guided Deep Surface Normal Estimation

Occlusion-Aware Depth Estimation with Adaptive Normal Constraints

Monocular Depth Estimation with Guidance of Surface Normal Map

Unsupervised Learning of Geometry with Edge-aware Depth-Normal Consistency

Virtual Normal: Enforcing Geometric Constraints for Accurate and Robust Depth Prediction.

Geometry-guided Multilevel RGBD Fusion for Surface Normal Estimation

Depth and Surface Normal Estimation from Monocular Images Using Regression on Deep Features and Hierarchical CRFs.

Adaptively Joint Pixel-wise Semantic Correlation in Surface Normal Estimation

Single Image Depth Estimation with Normal Guided Scale Invariant Deep Convolutional Fields

GeoNet++: Iterative Geometric Neural Network with Edge-Aware Refinement for Joint Depth and Surface Normal Estimation

Deep Surface Normal Estimation on the 2-Sphere with Confidence Guided Semantic Attention

NDDepth: Normal-Distance Assisted Monocular Depth Estimation

Metric3Dv2: A Versatile Monocular Geometric Foundation Model for Zero-shot Metric Depth and Surface Normal Estimation

Three-Filters-to-Normal$+$: Revisiting Discontinuity Discrimination in Depth-to-Normal Translation

Monocular Depth and Ego-motion Estimation with Scale Based on Superpixel and Normal Constraints

Three-Filters-to-Normal+: Revisiting Discontinuity Discrimination in Depth-to-Normal Translation

Deep Surface Normal Estimation with Hierarchical RGB-D Fusion

NDDepth: Normal-Distance Assisted Monocular Depth Estimation and Completion

NeuralGF: Unsupervised Point Normal Estimation by Learning Neural Gradient Function