On the first day of CP+ in Japan, Canon showcased its latest AI imaging technology:
From the press release
[Sometimes] inevitable problems with the image quality of photos. For example, noise that makes a photo look grainy, moiré that creates a pattern that shouldn’t be there, and image blur that’s caused by how a lens works. There were times when information was impossible to pull out. When using a wide-angle lens, due to optical characteristics, image quality at the corners of the frame tends to be blurred.
With the development of artificial intelligence technology and the application of deep learning technology, Canon, as a leading company with extensive knowledge of cameras and lenses, has taken on the development of deep learning image processing technology.
Canon’s deep learning technology for image processing, aimed at obtaining realistic photographs, is currently working on three areas: noise reduction, color interpolation and diffraction aberration correction.
Noise reduction that only removes noise while leaving details
You can capture the texture of smooth skin even when shooting at high sensitivities.
Image processing with color interpolation “Neural network Demosaic” for moire suppression
Improved color interpolation performance also improves resolution and color detail.
You can also get a higher resolution image by greatly suppressing the false color that occurs when shooting a striped shirt.
Aberration Correction
By correcting blur caused by aberration, object details are displayed clearly.
Comprehensive use of all three algorithms
Using all three directions of image correction produces a synergistic effect that significantly reduces aberrations and noise, providing unparalleled photographic expressiveness:
