Machine vision-based surface inspection system for rebar

Abstract

As the most basic component of a building structure, rebar is decisive for the safety and stability of a house building or infrastructure construction. With the advent of the era of industrial intelligence, it is particularly important to carry out intelligent inspection of rebar quality. The currently commonly used methods for detecting defects in rebar are usually inefficient and have poor detection accuracy. In this paper, based on machine vision technology, an efficient method of threaded wire head feature point screening is studied by collecting images of the surface of the rebar and analyzing the image information features, and the measurement of the wire head size and the detection and analysis of the size defects are completed. To further study the overall surface defect distribution of the rebar, this paper uses YOLOv3 target detection technology to finely locate and detect the surface defects of the rebar, and the speed and accuracy of the detection meet the automated inspection requirements.