Real-time implementation of super-resolution algorithms on plenoptic images

Real-time long-range target detection and tracking using conventional imaging is inefficient due to the optical limitations of such systems. In contrast, the plenoptic camera is efficient for target detection and tracking under deep turbulent conditions, as it can provide temporal and spatial information about the target. However, the inherent hardware design of a plenoptic camera provides a low spatial resolution. Image super- resolution (SR) algorithms can be implemented with the plenoptic system to increase the spatial resolution. Implementing image SR is challenging because low resolution (LR) images are devoid of detailed information about the target. This letter illustrates a complete setup for real-time plenoptic image capture and its super-resolution using an algorithmic approach. We develop an algorithm for SR and optimize it for real-time target detection. The proposed method is implemented in real time to superresolve LR plenoptic images by capturing a target image in turbulent atmospheric conditions placed 3.1 km away from the plenoptic camera. Each plenoptic sub-aperture image with 50 x 50 pixels is super-resolved eight times to 400 x 400 pixels. We illustrate an increase in the SR of the target without loss of information. The proposed scheme can be used in computer vision and remote sensing for object detection and to enhance the visual quality of the LR images.