Three-Dimensional Reconstruction Method of Oil Wear Particles Based on Binocular Stereo Microscopic Image

The three-dimensional reconstruction of oil wear particles can enhance understanding of their morphology and provide support for fault prediction and health management of aircraft engines. After acquiring binocular stereo microscopic images of abrasive grains in oil monitoring experiments, a model based on the principle of parallel binocular imaging can be established for 3D reconstruction without relying on precision instruments. Aiming at its limitation that the optical axis of the stereo microscope binocular camera is not parallel, which leads to a large amount of correction required in the stereo correction process, this paper proposes a three-dimensional reconstruction method based on the pyramid LK optical flow method. The comparative analysis of the results shows that this method can effectively reconstruct the three-dimensional morphology of oil abrasive particles within the allowable error range, and effectively reduce the errors introduced in the stereo correction process.