集成成像3D显示中微图像阵列转换研究

Integral imaging (II) is one of the most important true three-dimensional (3D) display methods, because it has no visual fatigue, requires no special glasses or coherent light. However, the existing II 3D contents are scarce, and the elemental image array (EIA) are only suitable for the II 3D display device whose specifications match with the EIA. Moreover, EIAs with different parameters cannot be transformed. These problems definitely hinder the development of II 3D display. This project plans to carry out the in-depth theoretical research and technological breakthroughs, and build up the optical models of the II pickup and display processes using different micro-lens arrays. We will explore the differences and internal relations between the different EIAs, and reveal the internal relevance mechanism of these EIAs. We will also explore the influence factors that induce the distortion of the reconstructed 3D images, and built up the representation model. We will set up the optical model for the EIAs’ transformation, deduce the corresponding relationships of the pixels in different EIAs, and finally realize the transformation of EIAs. To demonstrate the correctness of the transformation method and the fidelity feature of the reconstructed 3D image, the II 3D display test models and experimental platforms based on viewpoints and depth plans will be built up. The research work not only realizes the transformation of EIAs, but also provides the theoretical references for II 3D pickup, and promotes II to be put into practical application.

集成成像3D显示因无立体观看视疲劳、无需助视设备和相干光源等优点，是一种重要的真3D显示。但由于其片源匮乏，且某一参数的微图像阵列只能用于对应参数的集成成像3D显示设备，不同参数的微图像阵列间无法转换，阻碍了其快速发展。本项目拟从不同参数微透镜阵列的拍摄和显示光学模型入手，进行深入的理论研究和技术突破，探寻不同参数微图像阵列的区别和内在联系，从而揭示不同参数微图像阵列的内在关联机理；探索引起3D图像畸变的影响因素，建立3D图像的畸变表征模型；建立微图像阵列转换光学模型，推导不同参数微图像阵列的像素对应关系，获得不同参数微图像阵列的转换方法；建立基于视点和深度的集成成像3D显示验证模型及实验平台，验证不同参数微图像阵列转换方法的正确性，以及重建3D图像的全真性。本项目的研究工作不仅可以实现不同参数集成成像3D片源的有效转换，也能为集成成像3D拍摄提供理论指导，促进集成成像技术向实用化发展。

集成成像3D显示因无立体观看视疲劳、无需助视设备和相干光源等优点，是一种重要的真3D显示。但由于其片源匮乏，且某一参数的微图像阵列只能用于对应参数的集成成像3D显示设备，不同参数的微图像阵列间无法转换，阻碍了其快速发展。本项目从不同参数微透镜阵列的拍摄和显示光学模型入手，进行深入的理论研究和技术突破，建立了不同参数微图像阵列转换模型，提出了相应的像素映射算法。提出了光场重采样理论，深入分析了不同参数微图像阵列之间的像素映射关系，提出基于光场重采样的集成成像微图像阵列生成法，并验证了同一套拍摄系统生成不同参数微图像阵列的可行性。此外，所提出的基于卷积神经网络的集成成像片源生成方法、集成成像超大场景任意深度的3D重聚焦方法以及基于光场重采样的计算机二次拍摄法，实现了不同参数微图像阵列的转换以及片源的快速生成,从技术上解决集成成像3D片源的通用性问题。本项目的研究工作不仅可以提供有效的集成成像片源的生成方法，同时简化了集成成像片源的生成过程，提升了集成成像片源的生成效率，为不同参数和型号的集成成像3D显示器制作了丰富的集成成像3D片源，促进集成成像技术向实用化发展。