Research On Key Techniques Of Dark-field Confocal Microscopy With Axicon Pair Developed For Subsurface Defects Detection

High energy laser systems need a large number of optical elements which are expensive and difficult to process.Under the irradiation of high energy laser,this kind of optical elements is easy to produce high energy laser induced damage,which leads to the decrease of high energy laser system efficiency and the increase of operation cost.Studies have shown that subsurface defects(SSDs)are the main cause of high energy laser induced damage.Therefore,detecting SSDs in optical components nondestructively and obtaining high SNR SSD images are of great significance to ensure the healthy operation of high energy laser system.The main difficulties of detecting SSDs in optical elements nondestructively and obtaining SSD images with high SNR are as follows:The strong reflected light generated by the optical elements surface can inhibit the scattered light generated by the SSDs,and it is difficult to detect the SSDs by using bright-field confocal microscopy.The noise reduction performance of LRA-SVD algorithm is affected by the estimated noise variance,so the algorithm cannot meet the demand of dark-field confocal image noise reduction.Conventional image enhancement algorithms tend to cause excessive local image enhancement,insufficient image detail enhancement and background information amplification,which can not meet the demand of dark-field confocal image detail enhancement.In view of the above problems,this project carried out the research on key techniques of dark-field confocal microscopy with axicon pair developed for subsurface defects detection,and the main contents are as follows:(1)To solve the problem that it is difficult to detect SSDs with bright-field confocal microscopy due to the scattered light from SSDs is suppressed by the strong reflected light from the optical elements surface,a dark-field confocal microscopy imaging method based on axicon pair beam shaping and complementary aperture detecting is proposed.The beam shaping characteristics of axicons are analyzed based on the theory of scalar diffraction.It is proved that a pair of axicons can provide the ring light with narrow ring,high energy and large detecting aperture for dark-field confocal microscopy imaging system.The imaging characteristics of the dark-field confocal microscopy imaging system are analyzed.Based on the imaging characteristics of the system and the illumination energy of the ring light,the inner and outer ring radius ratio of the illuminated ring light is determined.The reflected light suppression characteristics of dark-field confocal microscopy imaging system are analyzed,and it is proved theoretically that the proposed method can suppress reflected light from surface and realize SSD detecting.(2)Aiming at the problem that the noise reduction performance of LRA-SVD algorithm is affected by the noise variance estimate,the algorithm cannot meet the noise reduction demand of dark-field confocal image,an improved noise reduction algorithm based on LRA-SVD is proposed.In the process of iterative denoising,the method noise is fed back into the denoised image and the noise estimation value,which avoids the influence of the noise variance estimate on the noise reduction performance of the algorithm by constructing a new noise image and improving the noise estimation model.The bright-field confocal image and the dark-field confocal image are used to verify the performance of the algorithm.The experimental results show that the proposed algorithm can both achieve image noise reduction and retain the details of the denoised image,and the performance of the algorithm is not affected by the noise variance estimation.(3)In order to solve the problem that conventional image enhancement algorithms tend to cause excessive local image enhancement,insufficient image detail enhancement and background information amplification,which cannot meet the demand of dark-field confocal image detail enhancement,the defect detail enhancement algorithm based on adaptive image fusion is proposed.The multi-luminance images are weighted and fused,which makes the enhanced image retain more detail information.A local edge-preserving filter is used to obtain a clearer edge in the enhanced image.Image entropy is used to obtain the optimal luminance transformation parameters of dark and bright regi ons,avoiding the problem of under-enhancement of dark region or over-enhancement of bright region caused by unreasonable selection of luminance transformation parameters.Dark-field confocal image enhancement results show that the proposed algorithm has excellent edge preserving ability and image brightness fidelity,and the local detail contrast of enhanced images is significantly improved.(4)The dark-field confocal microscopy imaging method is experimentally verified.Firstly,the dark-field confocal microscopy imaging experimental system based on axicon pair beam shaping is constructed.Then,the resolution characteristics,the surface reflected light noise suppression characteristics and the SSD imaging characteristics of the system are verified by experiments,and the engineering application ability of the dark-field confocal microscopy imaging method is tested by high energy laser Neodymium glass.The experimental results show that dark-field confocal can reduce the reflected light from neodymium glass surface by a factor of1.13×10~3,and realize the detection of neodymium glass SSDs.The proposed noise reduction algorithm and image enhancement algorit hm are used to process the dark-field confocal images of neodymium glass.The experimental results show that the SNR of the neodymium glass crack image increases from 2.92d B to 4.99 d B compared with that of the crack denoised image,and the SNR of the neodymium glass crack image increases from 2.92 d B to 6.81 d B compared with that of the crack enhanced image.