基于光度立体视觉三维重构算法的微观磨损形貌原位测量原理及方法

Image-based micro-wear morphology analysis plays important role in both wear mechanism and wear state analysis. This technique has been improved from 2D analysis to 3D analysis with the application of 3D microscopic imaging technology. However, for the maintenance and fault monitoring of equipment, the on-machine or in-situ measurement are more attractive than the advanced measurements described above. The reason lies in the fact that these expensive and advanced techniques still find their barriers in the application of in-machine or in-situ measurements. Aiming at this pain point, a new method based on the photometric stereo vision is introduced for the on-site and three-dimensional analysis of microscopic worn morphology. With a portable 2D image acquisition system, a 3D reconstruction method is constructed using multi-shadow 2D images. First, the calibration method is applied to perform the coordinate transform of the spatial points from the image coordinate to the camera coordinate. Second, the near-field point light source model is established and the surface 3D normal vector for each points is obtained using the photometric stereo vision algorithm. Finally, the surface depth and height for each points are calculated based on the FC algorithm, so far the 3D reconstruction of the wear surface micro-morphology is accomplished. Based on the above methods, a portable image acquisition system is developed to detect the plane and the curved surface respectively. Compared with the results of the laser scanning confocal microscopy, the current method shows acceptable precision with the height error of the feature points less than 4.2%, the cross-section curve error less than 10.9%, and the 3D surface error less than 12.9%. It can be generalized from above works that an effective technical means for in-situ and 3D analysis of worn surface of mechanical components.