A COARSE-TO-FINE MODEL FOR RAIL SURFACE DEFECT DETECTION

Abstract

Computer vision systems have attracted much attention in recent years for use in detecting surface defects on rails; however, accurate and efficient recognition of possible defects remains challenging due to the variations shown by defects and also noise. This paper proposes a coarse-to-fine model (CTFM) to identify defects at different scales. The model works on three scales from coarse to fine: subimage level, region level, and pixel level. At the subimage level, the background subtraction model exploits row consistency in the longitudinal direction, and strongly filters the defect-free range, leaving roughly identified subimages within which defects may exist. At the next level, the region extraction model, inspired by visual saliency models, locates definite defect regions using phase-only Fourier transforms. At the finest level, the pixel subtraction model uses pixel consistency to refine the shape of each defect. The proposed method is evaluated using Type-I and Type-II rail surface defect detection data sets and an actual rail line. The experimental results show that CTFM outperforms state-of-the art methods according to both the pixel-level index and the defect level index.

Existing System

Spectrum analysis, Bayesian, decision theory, and graph saliency models

Proposed System

The proposed method developed a coarse-to-fine model (CTFM) to detect defects from rail surface images. It includes three scales from coarse to fine: subimage, region, and pixel. The model first filters the defect-free subimages using a background subtraction model (BSM), which takes a transversal line as a modeling object and obtains a large-scale linear background model. It then locates the defect regions using a region extraction model (REM) that handles defects as saliency objects using phase-only Fourier transforms (POFTs). Third, CTFM refines defect shapes by a pixel subtraction model (PSM) that considers the pixel consistency along a longitudinal line and constructs small-scale pixel background models. Finally, CTFM fuses the three sets of results using a weighted geometric mean.

Architecture

BLOCK DIAGRAM