In situ reconstruction method for oil wear particles based on monocular multi-view imaging with planar mirror reflection

This paper presents a multi-view wear particle image acquisition method based on the principle of plane mirror reflection for wear particle 3D reconstruction. The LSTM 3D-R2N2 algorithm is used to reconstruct five wear particle images after image processing. The robustness of the method is validated using multiple types of wear particles. The reconstruction results are compared with data obtained from a surface profilometer. The results indicate that the proposed method is capable of identifying wear particle types and achieves high reconstruction accuracy across various particle types. Further validation through comparison with multi-view imaging confirms that the method maintains satisfactory reconstruction performance, even with only five input images. The experimental setup is simple, cost-effective, and demonstrates promising potential for industrial applications.