clepa Back CMOS-Based SWIR Camera

CMOS-Based SWIR Camera

Solving the low visibility challenge

 

Today, driver-assistance systems must deal with the challenges of low light and adverse weather conditions. In fact, most severe road accidents happen in these conditions: Even though the number of kilometres driven at night is substantially lower than during the daytime, more than half of all traffic fatalities occur after dark. 

 

Short-Wave Infrared (SWIR), refers to a specific wavelength range from 1000nm to 1600nm. SWIR allows for a number of applications to be performed that aren’t possible using visible light: a SWIR camera has a lower refractive coefficient, meaning that it is significantly less scattered and can perceive what standard cameras in the visible spectrum are not able to see.

 

Until now, SWIR sensing was extremely expensive because it is based on exotic materials compound, Indium Gallium Arsenide (InGaAs), and mainly used by defence and aerospace verticals to solve the low visibility challenge, which can afford the high price. TriEye’s patent-pending technology is able to overcome these obstacles and fabricate SWIR sensing on a CMOS-based (Complementary Metal-Oxide-Semiconductor) sensor, which reduces expenditure a thousand times compared to InGaAs. 

 

TriEye’s camera produces HD images of the driving scene, with incomparable efficacy under common low-visibility conditions. Delivering high-resolution image data to enable safer and more reliable assisted driving in low visibility conditions, better mapping of the car surroundings, and a higher object detection rate. This enables assisted driving applications such as emergency braking systems and pedestrian warning features to operate consistently, offering peak visibility day or night and in the most extreme weather.

 

SWIR image data can be processed with the same algorithms that were developed for regular cameras. Also, it is possible to use existing deep learning algorithms which simplifies the development process, saving significant time and resources, as the algorithms do not need to be developed from scratch which requires driving millions of miles physically and trillions of miles virtually.