基于等效曲面干涉图的大视角实时计算全息3D显示技术研究

As an ideal three-dimensional (3D) display technology, the computer-generated holography based on spatial light modulator (SLM) has broad application prospects in military, medical, scientific research, entertainment and other fields. However, there are still some problems hindering its further development. For example, the holograms are difficult to generate in real-time and the viewing angle of the reconstructed image is limited. The goals of this project aim to resolve the above problems. Firstly, according to the property of the field of view, the key factors affecting the viewing angle are studied, and the equivalent curved interferogram method is proposed. Therefore, the recording model of a large-angle hologram can be established to fundamentally expand the viewing angle. Moreover, the interferogram acceleration algorithms of single-object point and multi-object points are studied, and the GPU/CPU parallel processing computing platform is established. Through the effective combination of algorithm acceleration and hardware acceleration, we can realize the real-time calculating of the proposed hologram. Finally, the position compensation method based on phase shift factor and the multi-SLMs seamless splicing method based on holographic function screen are studied. Thus, high performance holographic reconstruction systems are established to obtain large viewing angle reconstructed images. Our research can realize large viewing angle real-time holographic display by simple operation, which has theoretical reference significance and application value in the field of the 3D display technology.

基于空间光调制器（SLM）的计算全息显示技术作为一种最理想的3D显示方式，在军事、医疗、科研和娱乐等众多领域具有广阔的应用前景。但是目前依然存在着全息图难以实时生成和再现像观看视角小等问题，阻碍着它的进一步发展。本项目为解决这些问题而展开研究：1）基于再现像的视区特性，研究影响再现像观看视角的关键因素，提出等效曲面干涉图法，建立大视角全息图记录模型，从根源上实现再现像观看视角的扩大；2）研究单物点和多物点干涉图加速算法，建立GPU/CPU并行处理计算平台，将算法加速和硬件加速有效结合，实现全息图的实时生成；3）研究基于相位平移因子的位置补偿方法和基于全息功能屏的多SLM无缝拼接方法，建立高性能全息再现系统，实现大视角全息显示。本项目的研究可实现大视角实时计算全息显示，并且整体操作简便，在该领域具有理论参考意义及应用价值。

基于空间光调制器（SLM）的计算全息显示技术作为一种最理想的3D显示方式，在军事、医疗、科研和娱乐等众多领域具有广阔的应用前景。但是目前依然存在着计算全息图（CGH）难以实时生成、再现像观看视角小、再现像上有斑点噪声等问题，阻碍着它的进一步发展。本项目为解决这些问题而展开研究：1）基于再现像的视区特性，研究影响再现像观看视角的关键因素，建立大视角全息图记录模型，被记录物体的尺寸为4.1mm，相比着传统方法，再现像视区扩大38倍；2）我们增加干涉图的使用效率，在保证再现像质量的前提下，减少浪费信息的记录，增加CGH的生成速度. 针对三个不同分辨率50 × 50、50 × 100和50 × 150的被记录物体，相比C-NLUT方法，所提方法的CGH生成时间分别减少52.75%、60.63% 和96.80%；3）通过像素分离和噪声平均，获得斑点噪声被抑制的大视角全息显示，相比GS算法，斑点噪声对比度降低54.55%；4）基于本项目所提出的大视角全息图生成方法，从多角度记录全息图，并通过设置特定旋转角度的振镜，利用简单光学系统，实现大视角的全息近眼显示。在本项目的支持下，课题组在国内外学术期刊上发表论文6 篇，申请专利 5项；参加本领域国内线下线上学术会议4次；先后培养青年教师和学生5人，组建了一支富有活力和创新性的全息显示研究团队。本项目的研究可实现高质量的计算全息显示，并且整体操作简便，在该领域具有理论参考意义及应用价值。