Computational Imaging for conformal imaging systems

传统的成像系统相对笨重，通常在一个近似长方体的成像系统中使用一个大透镜。这使得它们难以集成到许多平台中，特别是那些采用极端空气动力学表面（如鼻锥）的平台。该项目旨在开发计算成像系统，利用格拉斯哥大学开发的数学模型中的多孔径技术，演示保形表面的高分辨率成像，包括：(1)非对称傅里叶域多相机成像。下面的模型展示了这个主题的变化，展示了如何将多个矩形孔径组合起来产生各向同性的高分辨率成像。(2)计算成像，以减轻通过气动表面成像带来的像差。这项工作的一个关键目标是开发优化的成像系统，以检测关键对象，而不是检测图像。在这种情况下，机器学习等技术可以直接对像差、叠加的图像进行操作，以检测目标。

Conventional imaging systems are relatively bulky and typically employ a single large lens within an approximately cuboid imaging systems. This makes them difficult to integrate into many platforms, particularly those employing extreme aerodynamic surfaces such as nose cones. This project will aim to develop computational imaging systems that demonstrate high-resolution imaging from conformal surfaces using multi-aperture techniques developed as mathematical models at Glasgow, including: (1) Asymmetric Fourier-domain multicamera imaging. Variations of this theme are illustrated in the models below showing how multiple rectangular apertures can be combined to yield isotropic, high-resolution imaging. (2) Computational imaging for mitigation of aberrations introduced by imaging through aerodynamic surfaces. A key aim in this work is develop imaging systems optimised for detection of key objects rather than for detection of images. In this case, techniques such as machine learning can operate directly on aberrated, superimposed images for detection of targets.