Canon will launch software that uses AI to improve the quality of video captured by ultra-high-sensitivity cameras. It supports advanced surveillance by improving the visibility even in harsh environments such as long range and night time monitoring.

London, UK, 02 October 2023 – Canon Inc. announced today that the company is launching the “Video Enhancement Software Version 1.0” as software for the ultra-high-sensitivity camera series¹, which improves video visibility through noise reduction processing using AI in late January 2024.

In areas with extremely high security levels, such as seaports, public infrastructure facilities, and national borders, advanced surveillance systems are required to accurately capture subjects day and night. For night monitoring and long range surveillance, the noise generated in the dark environments or haze affects the clarity of the video and increases the risk of missing abnormalities. Canon has been strengthening its lineup of ultra-high-sensitivity camera series that can cope with capturing in such harsh environments.

The new product is software that makes it possible to further clarify video captured by these cameras by noise reduction processing using AI technology. By improving visibility, it achieves high-quality video surveillance and contributes to the immediate detection of abnormal situations. Canon will continue to strengthen and expand its advanced surveillance solutions in combination with the ultra-high-sensitivity camera series.

Proprietary deep learning image processing technology enables high-precision and high-speed noise reduction processing

The new product applies Canon's unique deep learning image processing technology developed by considering the characteristics of Canon's image sensor and optical system, as well as utilizing their vast image database and knowledge of image processing acquired over the course of many years. By having learned the noise characteristics of ultra-high-sensitivity camera series using this technology, it is possible to further reduce noise generated under low light environment, which cannot be avoided by the camera alone. In addition, Canon has made the deep learning models lightweight to reduce computational load and speed up the processing while maintaining the accuracy of model data accumulated through learning.

This proprietary technology realizes high-precision and high-speed noise reduction processing, thereby enabling more accurate and real-time understanding of subject information.