Image credit: global.canon
Canon has unveiled an AI-powered image processing technology to dramatically improve photos by addressing problems on noise, color, and lens aberration or blur.
The camera manufacturer says that its “cutting-edge technology” has been trained through the development of lenses and cameras, and images of “all imaginable subjects.”
With accumulated knowledge from a massive data set, the deep learning algorithm gets to understand how camera settings impact image quality as well as how to correct issues.
Canon says some problems are unavoidable when it comes to photo image quality. The company added they are “committed to the unavoidable ‘left behind in the camera’ principle of photography.”
These problems include noise that makes photos grainy, “moiré” patterns or false colors that shouldn’t exist, and image blurring caused by a lens.
The three specific issues Canon’s AI image processing tech covers are noise reduction, aberration diffraction correction, and color interpolation.
Noise removal or reduction isn’t new but when applied, details are typically lost and image quality suffers. With deep learning, Canon’s AI was able to remove noise with little image deterioration and still obtain crisp, high-definition results.
As for moiré colors, a new color interpolation processing capability called Neural Network Demosaic has been used to “strongly suppress” false colors while improving color reproducibility and resolution.
On problems that occur with lens issues like diffraction blur, Canon says it’s impossible to completely correct for all aberrations.
What its Neural Network lens Optimizer does is address image degradations to greatly enhance the amount of resolution a photo appears to have.
Canon did not specify whether the new AI image processing technology will be built into its cameras or made available in photo editing software.
