2017/07/24 by Nicola Strisciuglio, Strisciuglio, Nicola, George Azzopardi +3
Computer Science · Engineering · Medicine · #Computer Vision and Pattern Recognition (cs.CV) #Digital Imaging for Blood Diseases #FOS: Computer and information sciences #Image Processing Techniques and Applications #Industrial Vision Systems and Defect Detection #Retinal Imaging and Analysis
paper · pdf · doi:10.48550/arxiv.1707.07747
openalex publication_date 2017/07/24 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The detection of curvilinear structures is an important step for various\ncomputer vision applications, ranging from medical image analysis for\nsegmentation of blood vessels, to remote sensing for the identification of\nroads and rivers, and to biometrics and robotics, among others. %The visual\nsystem of the brain has remarkable abilities to detect curvilinear structures\nin noisy images. This is a nontrivial task especially for the detection of thin\nor incomplete curvilinear structures surrounded with noise. We propose a\ngeneral purpose curvilinear structure detector that uses the brain-inspired\ntrainable B-COSFIRE filters. It consists of four main steps, namely nonlinear\nfiltering with B-COSFIRE, thinning with non-maximum suppression, hysteresis\nthresholding and morphological closing. We demonstrate its effectiveness on a\ndata set of noisy images with cracked pavements, where we achieve\nstate-of-the-art results (F-measure=0.865). The proposed method can be employed\nin any computer vision methodology that requires the delineation of curvilinear\nand elongated structures.\n