Research on Feature Extraction Based on Time Series Images

The application of mold flux has greatly expanded and improved the continuous casting process and has become an indispensable metallurgical auxiliary material. In order to design a mold flux that is more convenient and better able to perform metallurgical functions, realize the automatic extraction technology of mold flux sequence image features, and study the relationship between temperature, time and phase distribution, it has become the top priority of current research. In this paper, the gray level co-occurrence matrix method is firstly used to analyze the crystallization and melting process of the mold powder in combination with relevant data and literature. Then use the image segmentation algorithm to intercept the central part of the image as the research object, and use the RGB color mode to reflect the color features through the color moment. Finally, the gray level co-occurrence matrix is used to describe the texture features, and the data is visualized and analyzed differently.