(B) Infrared

Infrared is electromagnetic radiation with wavelengths longer than visible light. The heat energy from the objects with a temperature above absolute zero is emitted as electromagnetic radiation, which is caused by the constant motion of molecules embodying heat. The electrons jump to higher energy band when they absorb energy by colliding with another. They can also release energy in the form of photons when falling to a lower energy band again. A hot molecule moves fast and generates higher frequencies (shorter wavelengths) of electromagnetic waves. Usually, the human eye cannot sense this radiation with infrared wavelengths, which can be measured by specific electronic sensors. Sensing the human body's infrared could deliver information such as body temperature, motion trajectory, etc. Two kinds of sensors are commonly utilized for this purpose: the passive infrared sensor (PIR) and the thermographic camera.

The electronic sensor PIR is designed to measure infrared (IR) light from objects. The term passive indicates that this sensor does not emit energy during the detection process. Instead, it detects the energy of infrared radiation from objects. It is widely used from motion detection to automatic lighting applications. In the field of HAR, PIR has been widely explored in the application of indoor positioning [153,154], device-free activity recognition [155,156], etc. The sensor is widely available in the market with low cost and low power consumption. The built system is privacy-secure and easy to deploy and maintain. However, a PIR sensor only detects general movement. It does not give information on who or what moved. For that purpose, a thermographic camera for imaging IR is required.

A thermographic camera generates an image by infrared radiation, which is different from a common camera sensing visible light. The objects with a temperature above absolute zero can be detected by the thermographic camera, and an object with higher temperature emits more radiation. Thus from the thermography, the temperature variations are also visible. For example, humans and other warm-blooded animals stand out very well against the environment, regardless of whether it is day or night. Thermography has been widely used in medical diagnosis, in the military, etc. In HAR applications, it has also been developed with image-processing algorithms for activity detection in residential spaces [157,158], muscle activity evaluation [159], respiration monitoring [160,161], etc. For detection in dark lighting conditions, namely in the work performed by Uddin et al. [162], the authors used the OpenPose framework for thermal images to check the possibility of body skeleton extraction. Their result shows that the thermal images can monitor humans in dark environments where the other typical RGB cameras fail. Although thermographic sensing could supply more detailed information on body action than the PIR approach, it suffers lightly from the cost and the computing load.
