With characteristics of low-cost and easy deployment, the distributed wireless pyroelectric infrared sensor network has attracted extensive interest, which aims to make it an alternate infrared video sensor in thermal biometric applications for tracking and identifying human targets. In these applications, effectively processing signals collected from sensors and extracting the features of different human targets has become crucial. This paper proposes the application of empirical mode decomposition and the Hilbert-Huang transform to extract features of moving human targets both in the time domain and the frequency domain. Moreover, the support vector machine is selected as the classifier. The experimental results demonstrate that by using this method the identification rates of multiple moving human targets are around 90%.

Pyroelectric infrared (PIR) sensor is low-cost, low-power small component commonly used only to trigger alarm in presence of humans or moving objects. PIR sensors are widely used as a simple, but reliable. In this paper, we present a novel pyroelectric sensor node, feature extraction method and human pyroelectric information fusion method to identify direction of multi-human motion, number of people passing and distance between the PIR sensor and people in an array of pyroelectric sensors. The pyroelectric sensor node is composed of two kinds of pyroelectric sensor modules. Each module has different function, so we can fuse the different function to accurately track human target. The approach has low computational requirement, so it is well-suited for limited-resources devices. Through the experiment, we can get the correct detection rate of the direction and number of people is more than 100%, the correct identification rate of the path is more than 78%.