多孔径成像是一种融合了仿生复眼视觉的新型成像方法, 具有小型化、大视场、高分辨率等多种优势, 但由于每个子孔径对应的单元图像分辨率过低, 导致其成像质量和视场角的提升十分有限。为了进一步提高成像分辨率和探测视场, 基于压缩感知理论设计随机编码模板, 并紧贴子孔径放置对入射光场进行调制, 通过单次曝光记录编码后的低分辨率单元图像阵列, 利用稀疏优化算法, 重构所有低分辨率单元图像获得超分辨率大视场图像。理论分析和仿真实验证明了该方法的有效性。该方法不仅能兼顾大视场高分辨率成像, 而且大大缩小系统等效焦距, 具有薄层结构, 体积小而重量轻, 可为微光机电一体化系统的研制设计提供借鉴。

Multi-aperture imaging is a new imaging method combining with compound eye concept, which has a small size, large field of view, high-resolution images reconstruction and other advantages. However, due to the low resolution of sub-images, the improvements for the image resolution and field of view are very limited. A novel imaging method which could achieve both super-resolution and large field of view was proposed. The random coded mask was designed based on the framework of compressive sensing and placed on each sub-aperture. Instead of directly imaging and converging on the image sensor, the incident light field of each sub-aperture would be modulated by the coded mask. Then, the random projections of the input object could be acquired by the low-dimension image sensor within a single exposure. Finally, the sparse representation-based optimization algorithm was applied to reconstruct super-resolution and large field of view images from all low-resolution sub-images, which had more object pixels than the number of pixels of the image sensor. Both the theoretical model and simulation results show the feasibility of the proposed method. Moreover, this method greatly reduces system equivalent focal length and has a thin structure, which can provide theoretical guidance for the design and application of the micro-optical electromechanical system.