Fast curvilinear structure extraction and delineation from line scan images using Multi-scale yin-yang discrete-point computing

Abstract

Extracting and delineating curvilinear structure is important for many applications. However, most of the existing algorithms cannot meet the real-time processing requirement. This paper presents a novel method for fast curvilinear structure extraction and delineation from TFDS-3 line-scan images using multi-scale yin-yang discrete-point computing, which can be divided into two stages: multi-scale yin-yang discrete-point sampling and multi-scale yin-yang discrete-point grouping. Multi-scale discrete-point sampling can effectually overcome the influences of interference from noise, textures and uneven illumination, greatly reduce the difficulty of centerline extraction, and tolerate small geometric transformations. By analyzing discrete-points' features in different belt-shaped regions of yin-yang discrete-point maps, multi-scale yin-yang discrete-point grouping produces a three-level detection system using feature points, line segments and centerlines, which can extract centerlines with their types quickly, smoothly and accurately from these belt-shaped regions, and achieve the final curvilinear structure results by clustering these centerlines with Gestalt rules.