CNET publishes notes from Image Sensors Europe conference being held in London these days. One of the articles talks about Swedish start-up called NocturnalVision developing spatio-temporal filtering of video stream to improve low-light performance. Henrik Malm, a professor at Sweden-based Lund University and co-founder of the start-up presented his technique. In a nutshell, it analyzes what's going on across each frame of an image--the spatial component--and what's going on from one frame to the next-the temporal component-to try to intelligently direct the noise reduction process.
This is the same technique that New Scientist talked about two months ago. So far Toyota spent 6M SEK supporting the research and has rights to the technology for automotive applications, Malm said, and NocturnalVision has rights elsewhere. The company is seeking venture capital and also is in talks with Sony Ericsson, Malm said.
Because of multi-frame nature of the noise reduction algorithm there's a delay of about six frames before the image processing can kick in. Also, the algorithm is quite resource-hungry, so the company uses Nvidia GForce 880GTX for parallel processing and is still limited by 6fps speed for VGA image, but Malm expects to improve it to 30fps with new hardware.
Another CNET note from the conference talks about efforts to standartize camera phone image quality and DxO Labs work in that direction.