Strip steel red rust defect detection method based on artificial intelligence

Abstract

The invention relates to a strip steel red rust defect detection method based on artificial intelligence, and the method comprises the steps: collecting an RGB image of strip steel, removing a background, obtaining a region of interest of the image, carrying out the graying processing of the region of interest, obtaining the gray-scale maps of channels R, G and B, extracting the texture features of the channel R through a gray-scale co-occurrence matrix method, obtaining a gray-scale map quality index of a channel, obtaining a gray-scale difference map of the channel at the same time, obtaining a red rust gray-scale feature map of a single channel according to the gray-scale map quality index and the gray-scale difference map, clustering pixel points of the feature map to obtain a connected domain, and obtaining an area and a contrast feature mean value of the connected domain; forming a feature matrix by the area, the contrast feature mean value and the value of the corresponding pixel point, taking the feature matrix as the input of a classifier, and outputting the defect degree of the red rust grey-scale map. According to the method, the feature map reflecting the red rust defect can be more accurately obtained, and then the red rust defect prediction is more accurate.

Description

technical field [0001] The invention relates to the fields of machine vision and image processing, in particular to an artificial intelligence-based red rust defect detection method for steel strips. Background technique [0002] In the metallurgical industry, during the heating process of hot-rolled strip steel before rolling, there will be special oxide scales that are strongly meshed with the base metal to form red rust. This defect generally appears in irregular strips or spear points along the entire width. On one side or on the upper and lower surfaces of the strip. This area is usually reddish after hot rolling, and rusts heavily to reddish brown, sometimes grainy and noticeably rougher than adjacent areas. [0003] At present, for the detection of red rust on hot-rolled strip steel, it mainly extracts and recognizes features through manual detection, image recognition and segmentation technology, and obtains recognition results. However, training and detection using...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a strip steel red rust defect detection method based on artificial intelligence, in order to solve the complex and inaccurate problems of the existing red rust degree detection method

Images