Arm’s Latest ISP Technology Redefines Visual Processing Abilities On A Digital Level

Do you know the visual processing in our retinas is controlled by over 2 million+ nerve fibres? It is an interesting fact especially for those who are trying to pull up the quality of digital visuals to the most real level. Till date, it has only proven to be a challenge for engineers who want to provide the viewers with most real-time viewing experience. Arm took it on itself to resolve the problem with its advanced Image Signal Processors (ISPs) – the Arm Mali-C52 and the Mali C-32 ISPs.

Fig. 1: Representational Image Showing Difference Between Image Quality With and Without Arm Technology

Apart from pulling up the level of viewing experience in the digital world, these ISPs will be addressing to the need raised by IoT devices for a more realistic viewing experience. Everything right from personal robots to smart home systems to IP cameras require a more efficient viewing experience for all.

The ISP technology is critical for the present day image technology that works with the aforementioned devices. The prime function of ISP is to yield the most precie and highest quality of image while extracting data from image sensor as well as while processing every single pixel especially at the points involving machine learning.

The new ISPs by Arm employ 25 processing steps for every single pixel out of which three are very important for bringing-in key differentiation with respect to output quality. These three key differentiators include color management, noise reduction, and High-Dynamic Range (HDR). The inculcation of Arm’s Iridix technology as well as other industry algorithms allows Mali-C32 and Mali-C52 ISPs to make a huge difference to these three aspects of pixels.

Applications

The Mali-C52 can be synced with two completely different optimizations – area and image quality. This kind of flexibility permits Arm’s silicon partners for using similar software and IP for plethora of use cases and products. The Mali-C32 can be specifically optimized for area in cost-sensitive, cost-sensitive mebedded vision devices like hobby drones or entry level access control.