Broadband laser-damage-resistant diffractive camera with high imaging quality

Wavefront coding imaging offers a promising path for laser protection of electro-optical imaging systems. However, current designs mainly focus on a single wavelength, since achieving diffractive achromats for dual requirements of optimal imaging and laser protection capabilities has been challenging. In addition, their laser protection effect is merely predicted rather than validated through actual laser-damage experiments. In this Letter, we report a broadband laser-damage-resistant diffractive camera with high imaging quality whose laser protection capability is verified in laser-damage test. A deep optics framework is constructed to co-design the learnable diffractive optical element (DOE) and image recovery neural network (NN) in an end-to-end manner. The NN is developed with two stages to maximize the restoration quality of the blurred images with variable laser glare. We fabricated the DOE using lithography. Simulation and experiment results both demonstrate that the diffractive camera can reduce the peak intensity of laser on sensor in spectrum of 473-688 nm by over 99%, thereby enhancing the laser-damage threshold by two orders of magnitude. Meanwhile, it also maintains high imaging quality of a typical peak signal-to-noise ratio exceeding 24 dB. The anti-laser camera renders great potential in imaging scenarios that may involve lasers, such as autopilots, drones, and security mentoring.