Research On Joint Optics-Image Design In Computational Imaging

Modern optoelectronic imaging devices require higher resolution,farther distance,and larger field of view for diversified applications.Traditional solutions adopt optical systems with larger relative apertures,longer focal lengths,more lenses,complex surfaces et al.However,disadvantages including complex structure,clumsy size and high cost are always accompanied,which makes it difficult to develop devices with small volume,light weight and low cost.A possible solution lies in image restoration algorithm to further optimize the residual aberration of optical system and improve the readability of image.Unfortunately,image restoration algorithm design is independent with the optical system design.For this reason,traditional optical imaging system is a little difficult to get the global optimal project of optical system design and image restoration algorithm design.Based on the global optimization principle in computational imaging theory,this study develops a joint optics-image design method.It takes simultaneously the two “nodes”,optical system design and image restoration algorithm,into consideration for optimization.The image restoration algorithm is introduced into the optical system design process for partial aberration correction.It produces a simplified optical system which has fewer lenses.Then the transmittance of optical system is improved,and the cost is reduced.Zernike aberration polynomials are first used to compute the PSFs and MTFs of optical aberration.Then the selective correction principle to optical aberration is established.Based on the principle and Zemax's programmable capability,an optics-image joint automatic optimization iteration is established for optical imaging system design.It makes it possible for a simplified optical system to achieve the imaging performance of a complex optical system.Major contents of this thesis are listed below:(1)MTFs of aberration are first computed with Zernike polynomials.According to the declining trend,the cutoff frequency and the zero-crossing condition,we evaluated the difficulty for image restoration algorithm to correct aberration.Furthermore,the selective correction principle to optical aberration is established by analyzing each aberrationdegraded image and the restored image.(2)Optics-image joint optimization utilizes Zemax's programmable capability to establish user defined operands(UDOP),thus the end-to-end image performance measures is used as the evaluation parameter in the automatic optimization iteration.Based on this approach,the joint optimization of optical systems with single lens or three lenses and image restoration algorithm is realized respectively.(3)Based on the study of spatial variation characteristics of PSF(SVPSF)and the optimal number of blocks for image restoration,the SVPSF sub-regional restoration is performed on the blurred image of the joint-design optical systems.Results demonstrate that restoration image contrast of the joint optimized single lens is improved by 5 times to the blurred image,and the three-lens restored image contrast is improved by 1.5 times.Bases on the fact that difference impacts of optical aberration on image degradation and degraded image recoverability,an optical transfer function/image deconvolution kernel is finally designed by the developed optics-image joint automatic optimization iterative approach.It is adaptive for restoration of blurred optical images.With the design kennel,the restored image quality of joint-designed three lenses is comparable to the traditionaldesigned six lenses,which proves that the optical joint design method can simplify the optical structure under the same imaging performance.It provides a new concept for highperformance SWa P computational imaging.Optical joint design method can be applied to the optical system design process of visible/infrared optoelectronic imaging devices,such as UAV monitoring systems,remote sensing mapping systems,medical equipment,machine vision lens,surveillance cameras,et al.