Research On Surface Defect Detection Method Of Dark-field Imaging Combined With Phase-shifting Interference Microsphere

The microsphere is an extremely important component form in many engineering fields,and its surface accuracy is one of the important indicators to measure its quality and reliability.High-precision detection methods for surface defects of small spheres usually have low detection efficiency and strict requirements on the detection environment.Therefore,the high efficiency,high precision,and high environmental adaptability of the detection methods for surface defects of small spheres have become the bottleneck problem of current research in this field.In order to meet the needs of high-precision detection of the surface of micro spheres in the fields of national defense,military,industrial manufacturing,etc.,and at the same time,in view of the main problems of existing methods,this thesis designs a detection method for surface defects of micro spheres based on vision combined with phase-shift interference.It has high detection accuracy while improving the timeliness and anti-interference ability of the detection system.Among them,the visual inspection part adopts the method of dark field imaging to make the defect highlight to the greatest extent,and make a preliminary judgment on the existence and size of the defect;the interferometric part adopts the method of multi-camera spatial phase shifting to carry out the three-dimensional shape of the defect.The depth measurement of the appearance improves the anti-interference ability of the system on the basis of ensuring the higher resolution of the interference image.Aiming at the problem of the inconsistency of the spatial positions of the cameras caused by the multi-camera system,this thesis proposes a method of autocollimator calibration combined with software registration to calibrate the camera's various angles and poses to ensure that the multiple interferograms collected by the camera are inconsistent.Each pixel has a one-to-one correspondence in space.At the same time,the spatial phase shift value of the polarizer is calibrated and compensated to minimize the error in data processing of the interferogram.Research and simulation are carried out on the interferogram data processing algorithm,and the characteristics of Flood Fill method,SRFNP method based on maximum phase gradient quality,and least squares iteration method in the circular area of unwrapping processing speed and antiinterference ability are compared and analyzed.The advantages and applicability of each algorithm in different application scenarios finally conclude that the SRFNP method is more suitable for unwrapping the interferogram in this experiment.In terms of wave surface fitting,the first four terms of Zernike polynomials are cut off by simulation software,which proves the effectiveness of Zernike polynomials in eliminating translation,tilt,and defocus in the X and Y directions.According to the principle of system detection,an experimental platform for rapid detection of surface defects of microspheres was built,and the software and hardware units of the detection system were developed to meet the design requirements of system resolution?10nm.The actual measurement of aerospace balls with a diameter of 3mm was carried out.The depth of a defect is0.940?m,which is basically the same as the result of 0.981?m measured by the laser confocal microscope,which meets the design requirements of defect detection accuracy ?50nm,and realizes high-precision and rapid detection of surface defects of tiny spheres.