Today many businesses employ fever detecting, tablet-based software technology to screen persons entering a building to reduce the spread of COVID-19 and ensure safety. The PDI-TK8 automatic thermal temperature scanner is designed for such non-contact uses and to improve safety versus hand-held temperature screeners at the doors. How does this innovative digital solution really work?
The computer vision software needs to perform fast and accurate face detection in this global use case. The science behind it is based on two things: identifying key facial features and getting the contours of detected faces. Knowing the distance to the person provides an important cue for face size normalization. This is why a device might ask the person it's scanning to step closer for an accurate reading. The software must also quickly handle variations in facial expressions, poses, and background lighting.
A novel camera imaging sensor known as time-of-flight (TOF) is used in the PDI-TK8. This imaging device and software for face detection rapidly provides both gray scale and depth information in real-time from a single sensor. Performance of face detection is improved from using both sets of information. This is useful to connect over a LAN to record staff temperatures, or it can even tie into building door access control systems.
FACE DETECTION (CASCADED CLASSIFIER)
Facial recognition software expedites and records staff temperatures.
Face detection on gray-scale images is done to efficiently select significant discriminative facial features with relatively high accuracy, and only negligible extra computation. The possible human eye regions are detected first by testing all the valley regions in the gray-level image. Then all the possible face regions which include the eyebrows, the iris, the nostril and the mouth corners are determined by the software.
The camera can obtain distance measurements for each pixel position on the image by measuring the light traveling time between emission and reception. Hint: Think of the pin art toy where you stick your face in it!
Face detection can be regarded as a specific case of object-class detection. In object-class detection, the task is to find the locations and sizes of all objects in an image that belong to a given class, in this case, the areas of the face. Facial recognition software allows you to upload a photo database of staff and record time and temperature.
HEAD SEGMENTATION (ACTIVE CONTOUR)
Based on a detected face region and using the additional depth image information, the head shown on the image is defined through an active contour method. Active contours are used for automatic object segmentation. The basic idea is this: a curve evolves to define the contours of the face.
Curve evolution is determined through an energy minimization approach. Since the depth variations within the face region are limited, an active contour model uses the depth variations in the interior and exterior of the contour to define the person’s head. Similarly, the PDI-TK8 can also be set to screen for mask use.
Automatic temperature readers enhance safety during COVID-19.
The PDI-TK8 device performs face recognition and measures body temperature in parallel using infrared induction thermometry combined with the TOF face recognition software. A 2D infrared array IR Sensor is used to measure the temp of the body. This automatic temperature reader enhances safety during COVID-19 with fast temperature scanning of employees, patients or visitors using an accurate thermal imaging camera. Using this camera provides non-contact human body temperature screening as a first line of defense to identify the presence of a fever as a symptom of COVID-19 or other contagious illnesses.