Collaborative On-Sensor Array Cameras

To address the issue of degraded image quality and generative reconstruction hallucinations caused by wavelength dependence in existing metal lenses, researchers proposed a collaborative metasurface array camera solution. This method involves jointly learning 100 million nanocolons to achieve end-to-end optimization under memory and computational limitations, while optimizing the large parameter array, the learned metatomic agents, and non-generative reconstruction methods with disparity and noise perception. Experiments show that this camera performs excellently in both simulation and all test scenarios, and its imaging performance is not affected by scene illumination spectra, achieving broadband imaging across the entire visible spectrum.