Abstract: The performance of the existing hot metal vortex image recognition system depends largely on the quality of the image. However, due to the interference of steel slag, light change and desulfurizer fog, the camera system faces some difficulties in obtaining the feature information of the vortex image, resulting in a large area of low quality in the image. In order to solve this problem, this paper proposed a method to identify and enhance vortex information in two-dimensional images based on the characteristics of hot metal and vortex. The method adopted an improved Gabor wavelet algorithm to extract the features and crack line information of vortex images. The image was processed by the feature processing unit to reduce the existence of false features and the quantization parameters of vortices were obtained by the vortex diameter recognition method. The experimental results show that, compared with the existing methods, this method reduces the pseudo-feature rate of vortex image information by 43.4 %, improves the accuracy of vortex diameter recognition by 3.6 percentage points, and realizes the enhancement and accurate recognition of vortex feature information on the free surface.

Key words: image recognition, vortex, grain line, gabor, vortex diameter