SKD: Keypoint Detection for Point Clouds using Saliency Estimation

TL;DR

This paper introduces SKD, a saliency-based keypoint detection method for point clouds that improves registration and reconstruction by combining geometric and descriptor responses, demonstrating significant performance gains on LIDAR datasets.

Contribution

The paper proposes a novel saliency estimation approach for keypoint detection in point clouds that integrates descriptor responses, enhancing robustness and accuracy.

Findings

Up to 50% improvement in matchability and repeatability on LIDAR datasets.

Higher inlier ratio and faster convergence in sparse matching tasks.

Effective integration of descriptor gradients for keypoint selection.

Abstract

We present SKD, a novel keypoint detector that uses saliency to determine the best candidates from a point cloud for tasks such as registration and reconstruction. The approach can be applied to any differentiable deep learning descriptor by using the gradients of that descriptor with respect to the 3D position of the input points as a measure of their saliency. The saliency is combined with the original descriptor and context information in a neural network, which is trained to learn robust keypoint candidates. The key intuition behind this approach is that keypoints are not extracted solely as a result of the geometry surrounding a point, but also take into account the descriptor's response. The approach was evaluated on two large LIDAR datasets - the Oxford RobotCar dataset and the KITTI dataset, where we obtain up to 50% improvement over the state-of-the-art in both matchability and repeatability. When performing sparse matching with the keypoints computed by our method we achieve a higher inlier ratio and faster convergence.