Selected work

Representative Papers

Rail Track Extraction from Rasterized Classified Point Clouds Using a Full-Resolution, Fully Convolutional Recurrent Neural Network

Jul 07, 2026

This study addresses the insufficient accuracy in automatic railway track extraction by proposing an end-to-end method based on a full-resolution fully convolutional recurrent neural network (FCRNN). The approach rasterizes classified point clouds and employs FCRNN for denoising and enhancing track contours, followed by morphological operations, centerline smoothing, and dynamic time warping (DTW) to reconstruct high-precision 3D track centerlines from 2D extraction results. Its novelty lies in being the first to apply FCRNN at full resolution for track extraction and relying solely on synthetic data for training. Experimental results demonstrate that the method automatically generates high-quality rail top surfaces and dual-track centerlines with minimal human intervention, significantly improving pixel-level accuracy and automation.

0 citationsRead paper

Efficient Computation of the Directional Extremal Boundary of a Union of Equal-Radius Circles

Mar 27, 2025

This paper addresses the efficient computation of the oriented extremal boundary—the contour formed by the farthest points along any given direction—of the union of equal-radius disks. Conventional approaches explicitly construct the union shape, suffering from high computational complexity and poor numerical robustness. To overcome these limitations, we propose the first plane-sweep algorithm based on circular arc event sorting, achieving an optimal $O(n log n)$ time complexity. By analyzing intersection points between disk pairs and their dominance relationships, our method directly extracts the critical boundary arcs without explicitly modeling the full union. Theoretical analysis confirms asymptotic optimality. Experimental results demonstrate that the algorithm achieves accurate boundary extraction under standard floating-point precision and outperforms naive geometric methods by over two orders of magnitude in runtime, while maintaining strong numerical robustness and scalability.

0 citationsRead paper

Recent publications

Latest Papers

Rail Track Extraction from Rasterized Classified Point Clouds Using a Full-Resolution, Fully Convolutional Recurrent Neural Network

Jul 07, 2026

This study addresses the insufficient accuracy in automatic railway track extraction by proposing an end-to-end method based on a full-resolution fully convolutional recurrent neural network (FCRNN). The approach rasterizes classified point clouds and employs FCRNN for denoising and enhancing track contours, followed by morphological operations, centerline smoothing, and dynamic time warping (DTW) to reconstruct high-precision 3D track centerlines from 2D extraction results. Its novelty lies in being the first to apply FCRNN at full resolution for track extraction and relying solely on synthetic data for training. Experimental results demonstrate that the method automatically generates high-quality rail top surfaces and dual-track centerlines with minimal human intervention, significantly improving pixel-level accuracy and automation.

0 citationsRead paper

Efficient Computation of the Directional Extremal Boundary of a Union of Equal-Radius Circles

Mar 27, 2025

This paper addresses the efficient computation of the oriented extremal boundary—the contour formed by the farthest points along any given direction—of the union of equal-radius disks. Conventional approaches explicitly construct the union shape, suffering from high computational complexity and poor numerical robustness. To overcome these limitations, we propose the first plane-sweep algorithm based on circular arc event sorting, achieving an optimal $O(n log n)$ time complexity. By analyzing intersection points between disk pairs and their dominance relationships, our method directly extracts the critical boundary arcs without explicitly modeling the full union. Theoretical analysis confirms asymptotic optimality. Experimental results demonstrate that the algorithm achieves accurate boundary extraction under standard floating-point precision and outperforms naive geometric methods by over two orders of magnitude in runtime, while maintaining strong numerical robustness and scalability.

0 citationsRead paper