High performance imaging in parallel cameras is a worldwide challenge in computational optics studies. However, the existing solutions are suffering from a fundamental contradiction between the field of view (FOV), resolution and bandwidth, in which system speed and FOV decrease as system scale increases. Inspired by the compound eyes of mantis shrimp and zoom cameras, here we break these bottlenecks by proposing a deep learning-based parallel (DLBP) camera, with an 8-μrad instantaneous FOV and 4 × computational zoom at 30 frames per second. Using the DLBP camera, the snapshot of 30-MPs images is captured at 30 fps, leading to orders-of-magnitude reductions in system complexity and costs. Instead of directly capturing photography with large scale, our interactive-zoom platform operates to enhance resolution using deep learning. The proposed end-to-end model mainly consists of multiple convolution layers, attention layers and deconvolution layer, which preserves more detailed information that the image reconstructs in real time compared with the famous super-resolution methods, and it can be applied to any similar system without any modification. Benefiting from computational zoom without any additional drive and optical component, the DLBP camera provides unprecedented-competitive advantages in improving zoom response time (~ 100 ×) over the comparison systems. Herein, with the experimental system described in this work, the DLBP camera provides a novel strategy to solve the inherent contradiction among FOV, resolution and bandwidth.

Imaging systems are widely used in many fields. However, there is an inherent compromise between field of view (FOV) and resolution. In this paper, we propose an optofluidic zoom system with increased FOV and less chromatic aberration, which can realize switching between large FOV and high resolution. The proposed system consists of a liquid prism, a zoom objective, an image sensor and image processing module, which can realize optical zoom and deflection. The proposed system achieves non-mechanical optical zoom from f = 40.5 mm to f = 84.0 mm. Besides, the angular resolution of zoom objective is up to 26"18 at f = 84.0 mm. The deflection range is ±10°, and the whole FOV of proposed system can reach up to 30.3°. The proposed system is compact and easy to machine. In addition, we reduce chromatic aberration produced by the liquid prism significantly. The proposed system can be used in monitor system, target tracking system, telescope system and so on.

Due to the relatively low optical power of a liquid lens, it is usually difficult to achieve a large zoom ratio and a high-resolution image simultaneously in an optofluidic zoom imaging system. We propose an electronically controlled optofluidic zoom imaging system combined with deep learning, which achieves a large continuous zoom change and a high-resolution image. The zoom system consists of an optofluidic zoom objective and an image-processing module. The proposed zoom system can achieve a large tunable focal length range from 4.0 mm to 31.3 mm. In the focal length range of 9.4 mm to 18.8 mm, the system can dynamically correct the aberrations by six electrowetting liquid lenses to ensure the image quality. In the focal length range of 4.0–9.4 mm and 18.8–31.3 mm, the optical power of a liquid lens is mainly used to enlarge the zoom ratio, and deep learning enables the proposed zoom system with improved image quality. The zoom ratio of the system reaches 7.8×, and the maximum field of view of the system can reach ~29°. The proposed zoom system has potential applications in camera, telescope and so on.

In this paper, we propose a real time, ultrahigh accuracy and full-field-of-view (RUF) algorithm for full field of view (FOV) imaging system. The proposed algorithm combines rough matching and precise matching method to stitch multiple images with the whole FOV in short time and high imaging quality. In order to verify real-time imaging effect of RUF algorithm, we also fabricate a multi-camera imaging system which includes 19 independent cameras. And the experiment result practically illustrates that full-FOV system can achieve good performances under a near-limiting FOV of 360° × 240° with low distortion, meanwhile, optical resolution reaches up to 95 megapixels. 100% registration-accuracy RUF algorithm for imaging in one second can be widely applied to any optical imaging engineering field with large FOV, such as remote sensing imaging, microscopy imaging, monitoring system engineering fields and so on.