Algorithm Research Of Spherical Optical Surface Defect Detection Algorithm

In the method of detecting spherical surface defects,the detection method of the microscopic dark field is a research hotspot,and the image processing algorithm is an important role of this defect detection method and is also the key to determining the speed and accuracy of detection.In this thesis,according to the detection requirements of the surface defect of the spherical optical element,the detection method of the microscopic dark field defect scattering is used as the research foundation,the image processing algorithms used in defect detection are studied,the main contents of this study are as follows:Firlstly,according to the detection principle and detection method of the surface defect of the spherical optical element,the overall scheme design of the defect detection algorithm of the spherical optical surface was completed.According to the characteristics of the defect position in the sub-image,sub-image classification algorithm was adopted to realize the sub-image classification.Secondly,The image registration algorithm of defect is studied.The registration algorithm based on image gray is improved.The edge feature of the image is used as the search path of the template.The pixel with the largest correlation is used as the registration position.At the same time,a single-scale Harris-SIFT algorithm is used to register the rotated images.The algorithm combines the Harris feature point detection algorithm with the SIFT description operator to eliminate the multi-scale of the traditional DoG-SIFT algorithm.The algorithms used in image synthesis are introduced in detail.The image interpolation algorithm is compared and the simulation experiments are performed.The interpolation algorithm suitable for defect images is determined.Thirdly,in the process of defect information extraction,by comparing the simulated noiseeliminated experiments,median filter could effectively eliminate the noise in the image wasproved,at the same time,this thesis combines the Canny edge detection algorithm and the Otsu threshold method to realize the image binarization.This method could effectively avoid defect fracture phenomenon.Then,using the minimum bounding rectangle algorithm,the classification of defect types and the extraction of defect evaluation information were completed.Fourthly,according to the shape characteristics of the spherical optical to be tested,the selection of the main devices such as the CCD camera,the lens,the illuminating light source and the wobble moving platform were completed,and the pixel calibration was completed by the electron microscope.The optics experiment platform was built with selected devices,and the defect on the surface of the spherical optical element were detected.The extraction and statistics of defect information were completed by using the algorithm proposed in this thesis.