Study On Near-eye Displays Technology Based On Human Visual System

Near-eye display can reconstruct a virtual scene by projecting the rendered light field information to vision system through a display device placed in a non-vision distance of the human eye.Currently,near-eye display technology is moving in a smaller,lighter,and thinner direction.At the same time,the content rendered by the near-eye display is also required to be more comfortable,more realistic,and more fluid.However,the existing near-eye display technology faces the challenges of excessive weight,excessive volume,and poor image quality.The weight and volume of the near-eye display will be effectively solved with the integration and miniaturization of the device.In the specific application scenes such as virtual reality and augmented reality,the main problem is the poor imaging quality which will result in weak stereoscopic feeling,the low resolution of the scene,the feeling of vertigo,visual fatigue and so on.This study focused on the three main factors that affect the image quality of near-eye display—three-dimensional scene reconstruction,system depth of field and monocular focusing cues.The research method included theoretical research,simulation and optical experiment and the theoretical basis were geometric optics theory,Fourier optics theory,Fourier slice theory,human eye physiological vision theory and four-dimensional light field theory.The spatial and frequency domain bandwidth of the four-dimensional light field in the near-eye display technology was studied.The relationship between the key parameters of the near-eye display and the retinal imaging quality was studied.The accurate real-time rendering capabilities for four-dimensional light field was enhanced,and the depth of reconstructed light field was expanded.The depth of field range and the relationship between adjustment errors and system parameters were constructed.In view of the problem of high accuracy reconstruction of light field,an algorithm of virtual viewpoints reconstruction via Fourier slice transformation was proposed based on the analysis of the EPI of the four-dimensional light field in the space and frequency domain.The new method effectively avoided information aliasing and boundary errors.Meanwhile,a dynamic priority and grid window strategy based on image depth information was constructed,and a fast boundary repair algorithm based on image content was proposed.Rendering time could theoretically be reduced by 3/4.In view of the problem of deep field expansion of the system in the near-eye display,a hybrid computational near-Eye light field display was constructed,and the optimization algorithm SPPM is proposed based on the minimization of the optical field matching error and the system light resolution.The results showed that the proposed optimization algorithm expanded the depth range of composite light sceneby by more than 50%,and the fluctuation range of the SSIM of visual perception imaging under different parameters were maintained within 4%.In view of the monocital focus characteristics in near-eye display,a general optical imaging model based on the visual characteristics of the human eye was constructed.It was the first time to explain the reason of accommodattion error derived from the conflict between spatial resolution imbalance and retinal blur effect.A method of evaluating retinal imaging quality of the near-eye display system based on the space loss rate was proposed,and the matching relationship between the human eye adjustment response error and the key optical parameters was constructed.The prediction results have the same trend as the validation experiment with high prediction accuracy.This work combining human visual characteristics and near-eye display optics will provide a basis for the development of a new generation of near-eye display devices.