Standards Development

Scope

This standard is used to measure object content of cast steel microstructures using fully automated image analysis from images produced by a microscope. The object content in the images must be clearly delineated by etching and recognized by experts.It includes the following contents:

(1) Requirements for the functions and parameters of the equipment in this standard;

(2) The accuracy evaluation method for intelligent analysis software; 

(3) The minimum accuracy requirements for intelligent analysis software;

(4) The operate procedures using intelligent analysis software.

Purpose

Casting is an important method for achieving efficient and cost-effective metal components., Castings are used extensively for manufacturing complex structural components. With the advancement of new industrialization, the demand for high-quality castings is steadily increasing, the performance of castings depends on their microstructure, Metallographic characterization is the most direct and effective method for obtaining microstructural information. Consequently, metallographic characterization serves as a vital link in quality control for casting and manufacturing process. There are several weaknesses in the traditional metallographic techniques. It relies heavily on expert experience, which can lead to errors, such as mistaking twinning grains for normal grains or measuring grain size with a low degree of precision. Secondly, due to limitations in human capacity, experts only examine in a limited field. Lastly, different experts introduce individual subjectivity in selecting the field of view, resulting in lower consistency in evaluation results.

The image analysis software is introduced to measure interest in microstructure with a high degree of precision. Traditional metallographic analysis software mainly employs threshold methods or watershed algorithms, requiring manual parameter setting, with relatively low recognition accuracy and robustness.

With the rapid integration and development of artificial intelligence technology and material genetic engineering, numerous intelligent metallographic detection software based on deep learning technology have emerged. Deep learning automatically extracts image features from large-scale data through back-propagation mechanisms, significantly improving the accuracy and robustness of image segmentation and recognition compared to traditional metallographic analysis technology. An increasing number of companies are applying this new technology to achieve high-quality metallographic characterization.