Wu Chensheng, Ko Jonathan, Davis Christopher C
Opt Express. 2016 May 30;24(11):11975-86. doi: 10.1364/OE.24.011975.
Under strong turbulence conditions, object's images can be severely distorted and become unrecognizable throughout the observing time. Conventional image restoring algorithms do not perform effectively in these circumstances due to the loss of good references on the object. We propose the use a plenoptic sensor as a light field camera to map a conventional camera image onto a cell image array in the image's sub-angular spaces. Accordingly, each cell image on the plenoptic sensor is equivalent to the image acquired by a sub-aperture of the imaging lens. The wavefront distortion over the lens aperture can be analyzed by comparing cell images in the plenoptic sensor. By using a modified "Laplacian" metric, we can identify a good cell image in a plenoptic image sequence. The good cell image corresponds with the time and sub-aperture area on the imaging lens where wavefront distortion becomes relatively and momentarily "flat". As a result, it will reveal the fundamental truths of the object that would be severely distorted on normal cameras. In this paper, we will introduce the underlying physics principles and mechanisms of our approach and experimentally demonstrate its effectiveness under strong turbulence conditions. In application, our approach can be used to provide a good reference for conventional image restoring approaches under strong turbulence conditions. This approach can also be used as an independent device to perform object recognition tasks through severe turbulence distortions.
在强湍流条件下，物体的图像在整个观测时间内可能会严重失真并变得无法识别。由于失去了关于物体的良好参考，传统的图像恢复算法在这些情况下无法有效执行。我们建议使用全光传感器作为光场相机，将传统相机图像映射到图像子角空间中的细胞图像阵列上。因此，全光传感器上的每个细胞图像相当于成像镜头子孔径获取的图像。通过比较全光传感器中的细胞图像，可以分析镜头孔径上的波前畸变。通过使用改进的“拉普拉斯”度量，我们可以在全光图像序列中识别出良好的细胞图像。良好的细胞图像对应于成像镜头上波前畸变相对且瞬间“平坦”的时间和子孔径区域。结果，它将揭示在普通相机上会严重失真的物体的基本真相。在本文中，我们将介绍我们方法的潜在物理原理和机制，并通过实验证明其在强湍流条件下的有效性。在应用中，我们的方法可用于为强湍流条件下的传统图像恢复方法提供良好的参考。这种方法还可以用作独立设备，通过严重的湍流畸变执行目标识别任务。
