Continuous casting billet surface defect binocular scanning and deep learning fusion recognition method and system

Abstract

The invention relates to a continuous casting billet surface defect binocular scanning and deep learning fusion recognition method and a continuous casting billet surface defect binocular scanning anddeep learning fusion recognition system. The method comprises the following steps: extracting a three-dimensional morphology image of the surface of a continuous casting billet by adopting binocularlaser scanning imaging; positioning a region of interest ROI according to the depth information; generating defect recognition candidate boxes. The real defect discrimination and the type recognitionare carried out on the defect area in the candidate box by adopting a full-connection neural network. For real defects, the semantic segmentation is carried out by adopting a full convolutional neuralnetwork. The three-dimensional quantitative morphological information of the defects can be accurately identified by fusing binocular laser three-dimensional scanning imaging and deep convolutional neural network target recognition and semantic segmentation methods. The method belongs to the field of continuous casting billet quality detection and control in ferrous metallurgy.

Description

technical field [0001] The invention belongs to the field of continuous casting slab quality detection and control in iron and steel metallurgy, and in particular relates to a method and system for fusion recognition of continuous casting slab surface defects by binocular scanning and deep learning. Background technique [0002] In the iron and steel metallurgical industry, continuous casting is the main technology for continuous solidification of molten steel and the source of raw materials for hot-steel direct rolling. It has been widely used by iron and steel enterprises since the 1990s, and its application technology and product quality have also developed rapidly. Billet surface quality monitoring technology plays a key role in the process of promoting the integrated production technology of hot feeding, hot charging and direct rolling, and gradually becomes an intermediate link of quality control, which can be used as a prerequisite for the realization of the integrated...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above problems, the present invention proposes a binocular scanning and deep learning fusion recognition method and system for surface defects of continuous casting slabs. Aiming at common harmful defects on the surface of continuous casting slabs under high temperature and high noise, through the fusion of binocular laser 3D scanning Imaging and deep convolutional neural network target recognition and semantic segmentation methods aim to accurately identify three-dimensional quantified morphological information of defects and realize reliable three-dimensional quantified detection

Images