An Optimized Union-Find Algorithm for Connected Components Labeling Using GPUs

TL;DR

This paper presents an optimized GPU-based union-find algorithm for efficiently labeling connected components in 2D images, significantly reducing runtime through a three-phase process.

Contribution

It introduces a novel three-phase GPU algorithm that improves efficiency in connected components labeling by minimizing atomic operations and focusing boundary analysis.

Findings

Over 1.3x speedup in average runtime

Effective reduction of atomic operations

Efficient boundary analysis implementation

Abstract

In this paper, we report an optimized union-find (UF) algorithm that can label the connected components on a 2D image efficiently by employing the GPU architecture. The proposed method contains three phases: UF-based local merge, boundary analysis, and link. The coarse labeling in local merge reduces the number atomic operations, while the boundary analysis only manages the pixels on the boundary of each block. Evaluation results showed that the proposed algorithm speed up the average running time by more than 1.3X.