Comparison between infrared-based and visible light-based facial recognition technology

Among different cooperative facial recognition systems, some use visible light-based (color) facial recognition technology. Such systems have trouble recognizing registered users under different light conditions. In a bright or dim environment, color-based facial recognition technology has a difficult time detecting objects. When color-based facial recognition systems capture images of users, the image clarity and quality is affected by surrounding illumination. In a relatively bright or dark environment, blurry, over-exposed or under-exposed images are often captured. Therefore, facial recognition systems using color-based recognition technology can only operate accurately in a controlled-light environment. In addition, the object and its background may affect color-based facial recognition technology as well. The color of the object and its background vary with respect to changing light conditions. Such variation is considered to be "noise" and greatly impacts the sensitivity and accuracy of color-based facial recognition.

PSP Security's facial recognition system (AccuFACE) uses infrared-based facial recognition technology. AccuFACE provides an active light source (infrared light) to illuminate the capturing object. Its video capturing module filters the ambient light and captures a clear infrared image of the user. Thus, changing light conditions or the object's background color has minimal impact to our captured images' quality. As a result, AccuFACE can operate accurately in a wide range of light conditions (from complete darkness to semi-outdoor environment). 

Visible Light Facial Recognition vs. IR Facial Recognition:  

Last month, we mentioned the mounting need for liveness detection for facial recognition. The Biomedical Signal Analysis Laboratory at Clarkson University, which focuses on processing and interpretation of signals arising from the human body, stresses the importance of liveness detection for all biometric systems, including facial recognition.

There are an increasing number of biometric applications in our lives. From access control to personal computers, biometric devices play a key and vital role to secure our physical and data security. However, as biometric device usage increases, methods of faking biometric characteristics for deceiving the devices also emerge. Therefore, liveness detection in biometric device is needed to avoid such imitations. For more information, please see:  http://people.clarkson.edu/projects/biosal/research/liveness.html