Theoretical And Experimental Study Of Surface Defects Imaging System For Steel Strips Based On Machine Vision

At present, it is of great significance for steel enterprises to control the surface quality ofcold rolled strips by online surface defect inspection system so as to improve product qualitylevel. The surface defects detection technology based on machine vision is fast, accurate andreliable, and thus has attracted much attention and has been widely used in industrial field. Thispaper thoroughly studied the design of cold rolled strips surface defects detection system basedon machine vision, and the key techniques of improving the system imaging quality andmulti-path image signal splicing, the main content and the results are as follows:1）A light scattering model used to describe the illumination intensity distribution on thesurface of the cold rolled strips was set up. Based on the surface optical properties of cold rolledstrips, and the BRDF lighting model theory, through the experimental analysis, the author got therelationship between the incident angle on the surface of cold rolled strips and light scatteringdistribution.2）Based on the analysis of the strip surface defect, the light path of the defect imaging wasoptimized. On the designed special experiment platform, the author did imaging light pathoptimization experiment for all kinds of defects, put forward a simple and reliable assessmentstandard for image quality, and according to the standard got the best imaging plan of all kinds ofdefects, provided basis for the light path design of the detection system and improvement of theperformance of the detection system.3）Image mosaicing based on camera calibration was completed. Through the analysis of theimaging model of a single camera, the author got the internal and external parameters of thecamera calibration, based on which the author set up binocular camera imaging model, andcompleted binocular system calibration with the aid of HALCON calibration board and functionlibrary. Based on the binocular imaging model the author put forward image mosaicing methodbased on the camera calibration, and verified its feasibility through experiments.4）The design and selection of the hardware of the whole detection system was completed.Based on small strip transmission experimental platform, according to the test requirements theauthor analyzed the imaging system parameters, did selection and installation debugging of thelight source, CCD, lens and the transmission mode, and set up the whole hardware system of thestrip surface defects detection.