[T9] Innovative Technology

[17E-T9-05] Exploration of Steam Oxide in Powder Metallurgy Using Image Recognition

*P. Wang^1,2, J. Fu¹, Y. Chen¹, T. Zeng¹, C. Lien², F. Lee³, C. Huang¹ (1.Feng Chia University, Taiwan, 2.Auroral Sinter Metals CO, LTD, Taiwan, 3.Brillant Knowbot Machines Co. Ltd., Taiwan)

Keywords:Steam Oxide, Image Recognition, Powder Metallurgy, Porosity

Recently, the use of AI-assisted methods for assessing the microstructures of metal powders has become increasingly popular. Especially in the context of powder metallurgy(PM), the thermal steam oxide process significantly influences mechanical properties such as hardness and porosity. the steam oxide process effectively reduces porosity and enhances hardness. Currently, porosity identification is primarily conducted by experts and cannot be quantified. Therefore, this study utilizes image recognition technology to apply various filters to enhance the regions of interest (ROI) by increasing image contrast and clarity. After image enhancement, grayscale recognition is used to binarize the images. This process clearly depicts the surface features of the specimens, facilitating further analysis and achieving both qualitative and quantitative results. Moreover, the comprehensive examination encompassed the correlation between the overall surface porosity and hardness of the test sample, which showed a difference in porosity between 6% and 10% for Rockwell hardness of 92 to 101 HRB. Utilizing image recognition, we explore the implications of these variations in hardness, porosity, and pore size.