3D Reconstruction Method of Image Based on Digital Microscope

  • Qinghua Xie


Most of the (3 dimensional, 3D) reconstruction requirements of micro scenes exist in industrial detection, and the scene requires real-time object reconstruction, which can get the surface information of objects quickly. For micro scenes, it is difficult to get the surface information. Mainly due to the shallow depth of field of the microscope, it is extremely easy to blur the image because the object surface is not in the focusing plane. Under the digital microscope, the images taken frame by frame are defocused images of most areas. The process of 3D reconstruction is to carry out geometric optical calculation through a single sheet or a few 2D images, obtain 3D information of objects by affine transformation, and complete 3D reconstruction. The feature of defocused image is that its complete information needs to be restored by a complete set of single-perspective defocused image sequences, and defocused image cannot complete the task of affine transformation due to lack of information. In this paper, 3D reconstruction method based on digital microscope images, the depth of the research in the current micro scene defocused image information recovery method, and the 3D reconstruction algorithm model, through the process of surface reconstruction based on point cloud data, the image registration, and filtering of double pass filter to the point cloud pretreatment, such as the noise and then the next sampling completed Delaunay triangulation based on planar projection and synthetic algorithm subdivision method, connect the point cloud to rest into the corners of another curved surface fitting, finally completed the 3D reconstruction experiments of sequence images collected, the experimental results show that the reconstructed surface basically conforms to the surface contour information of the selected objects.