It is possible to capture extremely weak visible light in the environment using a highly sensitive CMOS sensor, and then create a nearly realistic full-color image through AI depth denoising, color compensation, and a series of fine enhancement processes, ultimately realizing the core principle of low light full-color night vision.

The Niankang low light full-color night vision device can output full-color images in extremely dark environments of 0.003LUX. The entire process of driving at 120km/h is smooth and stable, without dragging or lagging, and can clearly observe road conditions and details of animal movements. By learning the noise distribution law through CNN/RNN, it directly generates denoised images, retaining more texture compared to traditional algorithms. The dual exposure technology achieves a dynamic range of 120dB+, solving the problem of overexposure/underexposure of human faces under backlighting and street lights, and optimizing local contrast.