Research On Recognition And Classification Of Steel Plate Head Shape

In plate rolling, normally, the plate out of the finishing mill will be fishtail-shaped or ear-shaped, the non-rectangular shape of the head and tail is caused by the different deformation in rolling. This part should be cut off to ensure the product is in rectangular size, which will reduce the yield. Therefore, the relevant control and optimization technology to improve the irregular plate shape becomes an important means of reducing head and tail cutting amount, improving the yield and saving energy.An issue in the control is how to recognize and classify the flat shapes.In this paper, the situation and image segmentation method are analyzed and studied based on the steel rolling. Then, I propose a conditional random field algorithm for image recognition, a statistical classification method for image classification, and thus obtain a piece of accurate information about the steel shape, provide a favorable basis for flat shape control in steel rolling.The main contents and results are as follows:First of all, the rolling field conditions were sub-fold, according to the complex conditions of the scene and harsh environment, select and install the image acquisition system, access to the steel head image.Secondly, several algorithms are used to identify the shape of the steel plate head through the collected images. The experimental results show that the use of conditional random field algorithm has a higher accuracy than others in the identification.Then, after the recognition of the image, the statistical classification method is used to classify the images. Compared with the k NN method, the proposed method has higher flexibility and classification accuracy.Finally, a software is designed to detect the plat shapes and complete the real-time detection of the steel plate shapes.Simulation shows that the proposed algorithm has a certain practicality, and it can effectively identify and classify the steel plate images collected in the field.