3-d Quantitative Detection Method Based On Laser Scanning Imaging For Surface Defects Of Strand

In recent years, with the development of hot charging and CC-DR technology, the strand quality was paid more attention than before. Even though the entire technical level in iron and steel industry had been improved greatly, the existing problem of strand surface defects and interior had been not resolved completely, therefore, NDT technology become the necessary link between steelmaking and rolling process.The 3-D quantitative detection method for defects of strand surface was studied in this paper, Based on machine vision technology, a laser scanning imaging system used area array CCD sensor was established, the main research contents are described briefly as follow:①A CCD image definition algorithm controlled by optical integral time was established through analyzing CCD sensor spectral response characteristic, a optimal feature extraction method of CCD laser line stripe was proposed though testing hot strand light intensity distribution and maximum intensity wavelength of the laser transmitter, using the same method, the area array CCD imaging light band and the filter center wavelength were determined.②The extraction method of laser stripe edge morphology on strand surface was established, the gray scale image and three-dimensional reconstruction image was accomplished too by controlling CCD frame scanning frequency. The measurement algorithm for defect depth was established base on the idea of laser triangulation range and key components spatial relations in system, according to the above theory a hardware system was designed.③The mapping relation between system-genertated gray-scale image and laser scanning distance matrix was established. According to the strand real-time moblie rate, the uniform accuracy detection algorithm of 2-D area for strand surface was established in horizontal orientation and vertical, it provide a theoretical basis for defects calculation and location on the strand.④Based on conventional image processing technology, a 2-D defect detection method that combined laser scanning image was improved, the pseudo-information (water film, iron oxide skin)on the strand surface can be eliminated by this method, in this system the anti-jamming capabilities and accuracy of ROI area positioning had been verified though the analisys of experiment results. ⑤The software of laser scanning system was developed and improved using VC++- Matlab joint programming and OpenCV technology, the system data was processed in real time using different processing modular,at the same time,some key technology included ROI segmentation on laser scanning image and edge detection and so on were studied in this system.⑥The foremost plement programs for industrial application was designed in this paper and had verifid the feasibility of the research method through experiment on site.This paper was founded by National Natural Science Foundation and shanghai Bao steel joint project (Grant No: 50974151). The relevant theoretical and experimental researches had been studied and the methods in this paper hace been experimentally verified.The 3-D quantitative detection method proposed in this paper have important theoretical and industrial application value.