The IEEE Signal Processing Magazine issued in March contains a paper "Radar Signal Processing for Elderly Fall Detection". This is probably the first survey paper of using radar techniques for human activity surveillance. The motivation of using radar for fall detection is that, the electromagnetic wave will have the so called Doppler effect under motions. Through capturing the Doppler effect signature signals using radar, signal processing techniques are used to extract the features and machine learning techniques are used for classification. It is evident that the combination of wavelet transformation and support vector machine lead to positive results.
There have been many times I heard people saying that intelligent surveillance is a sub-field of computer vision. However, I disagree with this idea, as I thought there should be many other types of sensors beyond vision that can be used for surveillance purpose. The field intelligent surveillance should be developed as an independent discipline by incorporating many other techniques, such as the techniques from electromagnetics. When the system have more than one type of sensors, the fusion of sensing information can be relied on workflow management systems.
?
?