Study Of Singlet Lens Computational Imaging Technology Based On Circular-symmetric Aberrations And Its Applications

Modern high-performance optoelectronic systems pursue high-performance imaging and lightweight systems,especially in aerospace,intelligent imaging,and mobile medical fields.The trend of lightening optical systems is to adopt imaging methods combined with singlet optics and computational imaging algorithm.However,the existing singlet optics combined with computational imaging still have the problems:(1)The traditional aberration merit-function does not consider the correlation with the computational imaging algorithm,and cannot optimize the imaging quality of the combination of the two.Therefore,it is necessary to construct the aberration merit-function of a singlet optics suitable for the computational imaging algorithm.(2)Since the singlet optics needs to achieve the imaging quality of the multi-lens complex optical system,its surfaces are more complicated than that of traditional optical elements.Therefore,it is necessary to study a testing method with high flexibility,high efficiency and low cost,so as to solve the problem of surfaces testing of customized designed singlet optics.Based on the above research background and existing problems,the main research contents of this thesis include:In order to solve the problem of constructing the aberration merit-function of the singlet computational imaging system,a singlet lens aberration balance method based on the circular-domain symmetric merit-function is proposed.According to the mathematical model of the imaging algorithm,the corresponding singlet lens design requirements are firstly established.The corresponding aberration requirements are that the asymmetric aberration terms are completely eliminated,and the symmetrical aberration term is allowed to be partially retained.Then the Zernike polynomial is used to transform it into a control function for 36 Zernike polynomial coefficients.In actual balance process,the 36 coefficients were parameterized with the surface parameters(r,k,n,?_i)of the singlet lens as variables,and then the explicit functions of the coefficients and surface paremeters are constructed.The optimization goals of asymmetric and symmetric aberration functions are set and the local optimal solutions are solved by the variable-step least square method.Using the proposed aberrations balance method,the singlet microscope and the singlet space telescope were customized according to the computational imaging algorithm,and the feasibility of the proposed aberrations balance method in the finite distance conjugate distance and the infinite distance conjugate distance systems is verified.In order to solve the problem of surface testing of customized singlet lens,a dynamic angle compensation method for complex surfaces interferometry is proposed.Based on the principle of angle compensation,the dynamic programmable Tip/Tilt mirror is proposed as the dynamic angle compensation element,and the active and flexible angle compensation method is realized.This method can adaptively plan the compensation angle and deflective route of the Tip/Tilt mirror and adaptively realize high-precision detection of complex surfaces with multiple degrees of freedom.A unified error model based on the Tip/Tilt mirror was established to avoid the need for calibration experiments every time the tilt angle was changed,which effectively saved the testing time.Finally,it is proved by experiments and data,where the surface testing accuracy of this method is better than?/10,and the maximum measurable surface slope angle is 2.8°.In addition,the built interferometry experimental setup is used to test the surface of the customized singlet lens,and the good compatibility would also reduce the processing/inspection cost.In order to verify the feasibility of the imaging method proposed in this paper,singlet lens computational imaging experiments based on deep learning were carried out.Microscope imaging experiments and space telescope imaging simulations both proved the feasibility of the proposed singlet lens computational imaging method,that is,the residual symmetrical aberrations can be effectively eliminated by computational imaging algorithms.It lays a foundation for the practical application of lightweight and portable singlet lens computational imaging system.Finally,in order to verify the effectiveness of the customized singlet computational imaging system proposed in this paper in practical application,a set of portable multi-functional biomedical microscopy imaging based on deep learning was designed and built(multi-spectral imaging and virtual phase contrast imaging).Through the imaging experiment of the USAF-1951 resolution target,it is proved that the imaging resolution of the proposed method(417nm@1.38?m)is better than that of the scientific commercial microscope objective lens(417nm@1.95?m).Through imaging experiments on tumor pathological slices(unstained),it is verified that the virtual phase contrast singlet microscopy imaging in this paper can achieve the effect of traditional commercial phase contrast microscopes,and its SSIM is 0.75.