Font Size: a A A

Research On The Surface Depth Measurement Method Based On Monocular Microscopy Objects

Posted on:2024-09-10Degree:MasterType:Thesis
Country:ChinaCandidate:X J YuFull Text:PDF
GTID:2568307181951129Subject:Electronic Information (Computer Technology) (Professional Degree)
Abstract/Summary:PDF Full Text Request
In recent years,precise measurement and precision manufacturing have become an inevitable future development trend due to the rapid development of industrial manufacturing and modern science and technology.Benefiting from the rapid development of artificial intelligence and machine vision technology,image-based 3D reconstruction has emerged as a hotspot for research in the field of machine vision.Compared with traditional 3D models obtained by using modeling software or laser scanning methods,image-based 3D reconstruction offers a higher degree of automation and fewer constraints.Numerous image-based 3D reconstruction algorithms have been applied to imaging large scenes with rich texture details.However,in microscopy applications,observing specimen mainly relies on two-dimensional,and it is difficult to observe the three-dimensional features of the specimen.Therefore,microscopeoriented 3D technology is of great importance.The shallow depth of field and small field of view of microscopes limit the ability to obtain three-dimensional morphological information about tiny objects.Traditional stereo microscope is to imitate the human eye through binoculars to acquire stereo image pairs,the limitation is that the optical path angle is limited only for microscopic systems with low magnification.Monocular microscope can only acquire depth information by moving the microscope or the target object when capturing images,but which is limited by the small field of view of the microscope and often requires precision operation.Therefore,in this paper,we investigate the above-mentioned problems:(1)For the problem of difficult microscope image acquisition.A monocular stereo vision system based on the symmetric refraction of tilted flat glass is proposed.Since the monocular straight microscope used has a deep depth of field and a long working distance,there is enough space for operation.Firstly,the tilted plate glass is placed between the target object and the objective lens,and the displacement produced by refraction is equivalent to the change of viewpoint;then,by symmetrically rotating the tilted plate glass left and right,the stereo image pair can be obtained;finally,the virtual binocular stereo vision method is used to realize the depth measurement of the object.(2)For the problem of refractive distortion introduced by the tilted plate glass.A symmetric refraction depth-based method is proposed to reduce the effect of refraction.Through the obtained image pairs,the SIFT algorithm is used to obtain a match between symmetric refraction stereo image pairs,and then the RANSAC algorithm is used to reject the mismatched points.Combining the information on the location of imaging points generated by the refractive optical path of the tilted flat glass,a flat plate refraction model is established.In the case that the plate glass intrinsic parameters and tilt angle are determined,the depth information of the object surface can be recovered according to the geometric position relationship between the direct imaging point and the refraction point to reduce the error caused by refraction.(3)In response to the problem of view loss caused by the tilted tablet glass under the high mirror microscope.A multi-view system with rotation refraction was proposed.When the high magnification of the microscope is high,it is difficult to control the image to produce an effective overlapping area when collecting pictures.Therefore,it is no longer limited to two perspectives when collecting,but instead collects the image sequence under multi-view by rotating tilt flat glass.Use an incremental SFM to reconstruct the target object.Use the beam flat to optimize the 3D model to improve the reconstruction accuracy.
Keywords/Search Tags:Monocular microscope stereo vision, Tilted plate glass refraction, Surface depth measurement, Multi-view reconstruction
PDF Full Text Request
Related items