Automated computer-vision-based defect detection has received much attention with the increasing surface quality assurance demands for the industrial manufacturing of flat steels. This article attempts to present a comprehensive survey on surface defect detection technologies by reviewing about 120 publications over the last two decades for three typical flat steel products of con-casting slabs and hot- and cold-rolled steel strips. According to the nature of algorithms as well as image features, the existing methodologies are categorized into four groups: statistical, spectral, model-based, and machine learning. These works are summarized in this review to enable easy referral to suitable methods for diverse application scenarios in steel mills. Realization recommendations and future research trends are also addressed at an abstract level.

/pdf/automated-visual-defect-detection-for-flat-steel-surface-a-3wjhcxn3aa.pdf

Automated Visual Defect Detection for Flat Steel Surface: A Survey

Surface defect classification of steels with a new semi-supervised learning method

A deep-learning-based approach for fast and robust steel surface defects classification

Defect detection of hot rolled steels with a new object detection framework called classification priority network

The automation and intellectualization of the manufacturing processes in the iron and steel industry needs the strong support of inspection technologies, which play an important role in the field of quality control. At present, visual inspection technology based on image processing has an absolute advantage because of its intuitive nature, convenience, and efficiency. A major breakthrough in this field can be achieved if sufficient research regarding visual inspection technologies is undertaken. Therefore, the purpose of this article is to study the latest developments in steel inspection relating to the detected object, system hardware, and system software, existing problems of current inspection technologies, and future research directions. The paper mainly focuses on the research status and trends of inspection technology. The network framework based on deep learning provides space for the development of end-to-end mode inspection technology, which would greatly promote the implementation of intelligent manufacturing.

https://www.mdpi.com/2076-3417/8/11/2195/pdf

Research Progress of Visual Inspection Technology of Steel Products—A Review

Application of RNAMlet to surface defect identification of steels

2D Detection with gray level images is popularly employed in current surface inspection systems of steel strips, which lead to high false detection rate because of interference of pseudo-defects, such as iron oxide, water stains, and grease. Therefore, an online detection technique of 3D defects based on photometric stereo was developed in this paper. The acquired 3D information is utilized to distinguish real 3D defects from pseudo-defects, which can significant improve the performance of surface inspection system. The designed inspection system was equipped with lasers and linear CCD camera to scan the moving surface of steel strips, referred as the photometric stereo laser scanning system. Samples of hot and cold-rolled steels, with pseudo and real defects such as roll imprints, cracks and scratches on surface, were scanned and showed obvious distinguishment.

Online Detection Technique of 3D Defects for Steel Strips Based on Photometric Stereo

The invention relates to a surface roughness measurement method and a realization device. The method is characterized by using a high-resolution microscopic camera which is equipped with a long objectdistance microscopic lens to shoot the surface image of a measured object when a single point light source irradiates, wherein after n point light sources are successively lightened, the camera can shoot n surface images I1-In, and the n is greater than or equal to 3; using a high-light black ball to calibrate the direction of the light source and acquiring the unit direction vector L1-Ln of allthe light sources and the transformation ratio coefficient f of a world coordinate (x, y, z) and an image coordinate (u, v); calculating the image coordinate value W(u,v) of a surface depth through the unit direction vector L1-Ln of the light source and the two-dimensional image I1-In of the surface of the measured object; acquiring the world coordinate value Z(x, y) of the surface depth of the measured object through the transformation ratio coefficient f; selecting the appropriate sampling length lr, and sampling the Z(x, y) to acquire a contour curve z(x); using a gaussian filter to filterthe contour curve z(x) so as to obtain a gaussian center line w(x), and separating a roughness profile r(x); and calculating the arithmetic average deviation value of the roughness profile r(x) and taking as the measured value of roughness.

Surface roughness measurement method and device

The invention relates to a method for achieving three-dimensional reconstruction of the steel plate surface A red monochromatic light source, a green monochromatic light source and a blue monochromatic light source light the steel plate surface at different angles, the angle between the incident light and the steel plate rolling speed direction ranges from -50 degrees to 50 degrees, a color CCD video camera perpendicular to the steel plate surface shoots the steel plate surface area irradiated by the light sources, a red channel, a green channel and a blue channel of a color image are separated, and three images obtained under independent lighting of the red light source, the green light source and the blue light source can be obtained According to the method, the complexity and the number of image pixels are in the linear relation, and the overall optimization algorithm is in the square relation, so that time consumption of the method is largely reduced, and the error of the method is slightly smaller than that of the overall optimization algorithm

Method for achieving three-dimensional reconstruction of steel plate surface

The invention provides a medium-thickness plate optimal shearing control method based on surface quality on-line evaluation, and belongs to the field of medium-thickness plate quality control. The method comprises the following steps of building an online detection system for defects on the surface of a medium-thickness plate, constructing a hierarchical structure model, outputting a grading result of the quality of the surface of the medium-thickness plate and determining a optimal medium-thickness plate blank shearing strategy, and realizing intelligent control of medium-thickness plate optimal shearing based on the surface quality on-line evaluation. According to the method, the waste caused by sizing and shearing of a medium-thickness plate blank is overcome, the conversion from artificial subjective control of shearing to scientific model control of shearing is realized, the shearing result is more scientific, and a good foundation is laid for efficient production and improvementof the quality of terminal products.

Wang Lei

Papers

Application of RNAMlet to surface defect identification of steels

Online Detection Technique of 3D Defects for Steel Strips Based on Photometric Stereo

Surface roughness measurement method and device

Method for achieving three-dimensional reconstruction of steel plate surface

Medium-thickness plate optimal shearing control method based on surface quality on-line evaluation