Image analysis of crystallizer flux crystal sequence based on OCR model

Because of the oxidizability of molten steel and the conduction loss of heat in the metallurgical process, it will cause greater interference to the process of metal smelting. The performance of mold flux directly affects the casting speed and the quality of continuous casting production. Therefore, it is of great significance to study the dynamic characteristics of the melting and crystallization process of mold flux during continuous casting. In this paper, the automatic feature extraction technology of sequential images is studied to solve the problem that the high temperature, transient fluid flow, complex phase change and chemical reaction in the mold and the opacity of the mold wall bring difficulties to the batch direct observation of the crystallization behavior of the mold flux in the metal smelting process. The temperature-time relationship model was obtained based on the OCR model, and the relationship between crystallization agent temperature, melting rate and crystallization rate was obtained by using the nonlinear regression model to fit the characteristic change trend of the image corresponding to the heating time.