Research On Computational Imaging Technology Of Simple Optical System

In order to meet the demand for obtaining high-quality images,modern optical imaging systems are designed to be very complex,which brings many problems such as large size,heavy weight and high price.While meeting the high image quality of the system,the realization of low cost and miniaturization of the system has important significance and wide application prospects in the fields of consumption,security,remote sensing and medical treatment.Computational imaging technology for simple optical systems is a rising technology using image processing methods to simplify optical systems.This technology relaxes the requirements of imaging quality in the front-end optical system design,and uses image restoration algorithm to remove the residual aberrations in the back-end,which simplifies the system and achieves the goal of low cost and miniaturization of the front-end optical system while ensuring high image quality.The thesis takes the computational imaging technology of simple optical systems as the research topic,and carries out the following work around the three important components of the technology: optical image restoration algorithm,point spread function(PSF)acquisition and optical/algorithmic co-design method:1)Research on image restoration algorithms.The factors causing the degradation of the optical system are analyzed in detail,and are classified into two types: causing image blurring and deformation.The two types of image degradation models are established.According to the model and the current advanced image deconvolution algorithm,a spatially varying cross-channel image restoration algorithm suitable for simple optical systems is proposed.The algorithm first corrects the lateral color aberration and distortion of the image,which ensures the validity of the cross-channel prior;then uses the cross-channel deconvolution algorithm to restore the blurred image patches,which effectively reduces the chromatic aberration that is difficult to correct in the optical system design and restores the high-quality image.2)Research on PSF acquisition.The concept and characteristics of PSF are introduced,and the importance of accurate PSF acquisition to the quality of restored images is explained.The structure and the wide-band characteristic of the color image sensor,which is a key and easily neglected component for PSF acquisition,are analyzed.The wide-band characteristic of PSF and the limitations of two PSF acquisition methods,PSF measurement and PSF simulation,are discussed.Then,the idea of using the measured PSF to modify the PSF simulation to obtain the wide-spectrum PSF is proposed.Based on this idea,two methods of wide-spectrum PSF acquisition methods for real optical systems are proposed.One is calculating the wide-spectrum PSF according to the weighted sum of the measured object spectrum and single-wavelength PSFs,which can accurately estimate the wide-spectrum PSF;for the scenes where real object spectra are complex,the other generates an optimal wide-spectrum PSF which can reduce the blur caused by all the wavelengths in the color channel of the sensor.The practical experiments show that both methods can reduce the degree of blur caused by the simple optical system and hardly introduce artifacts,which ensure the stability of image restoration,and obviously have better performance than the current advanced PSF acquisition methods.3)Research on optical/algorithmic co-design method.An optical/algorithmic co-design method for simple optical systems is proposed.Combining with the image restoration algorithm proposed above,in optical system design,the constraint of chromatic aberration is relaxed and the optimization weight of the green channel is increased to make the green channel image sharp,which simplifies the system and ensures the quality of the restored image.Using this method,a simple optical system with large field and high image quality which has only two lenses and a simple telephoto optical system consisting of 7 spherical lenses with low telephoto ratio are designed.Experiments show that the blurred images produced by the two systems can obtain high quality images after image restoration.