When there are few photons to capture, then noise is always a problem. Our AI-based approach achieves a new level of image denoising, and unlike other high-performance denoisers, works in real-time. For the first time, we're enabling real-time imaging in extremely low lighting conditions, in full colour.
Traditional image processors discard a critical portion of sensor data, limiting image quality. Images are compressed and smoothed, leaving missing opportunities to use temporal information between successive frames. By using raw Bayer image sensor output, our AI denoiser can identify and remove even more noise.
With the power constraints of real-world electronic devices, computing performance is critical. Our denoiser has been developed to run on a choice of cost effective, low power, neural networks at the edge.
Reducing image noise delivers a whole new degree of freedom for system designers. Noise removal can reduce the cost and size of image sensors, reduce the energy wasted in illumination, increase the frame rate, or improve the dynamic range.
To get an idea of how much further your sensor can go with our software, send us some sample data and we'll provide you with a quick, direct comparison. We suggest providing samples of your raw Bayer sensor data under the most challenging lighting conditions, and we'll send you back a demo of what we can achieve with your current image sensor.