Abstract:Ferrograph analysis technology has been adopted for determining the wear status of mechanical equipment,but it can only provide two-dimensional image of wear particles,resulting in incomplete characterization of wear particles.To accurately reconstruct the three-dimensional(3D) morphology of wear particles,a 3D reconstruction method by introducing photometric stereo vision into the ferrography analysis process was proposed.The wear particle regions are firstly segmented from the image background using the the maximal variance between clusters method based on the illumination spatial variation of the full-light images.Then,in order to eliminate the reconstruction error of wear particle morphology caused by the illumination intensity differences of different LEDs,the ideal imaging brightness of background regions is defined by the plane shape and Lambertian reflection model,and then the brightness distribution of each photometric image sequence is corrected.Finally,3D wear particle surfaces are reconstructed based on the principle of photometric stereo vision from the corrected photometric image sequences.With different types of wear particles,the proposed method is compared with the confocal laser scanning microscope.The reconstruction results of the proposed method were compared with the measurement results of confocal laser scanning microscope using different types of wear particles as test samples.The result shows that the parameter error of the reconstructed wear particle morphology is less than 15%,indicating that the proposed method can accurately reconstruct the3D morphology of the wear particles.