Multiple Human Monitoring with Wireless Fiber-Optic Multimedia Sensor Networks

This paper presents a binary compressive sensing based fiber-optic sensor system for human monitoring. Fiber-optic sensors are flexible and convenient to measure pressure information of humans. Such a nature enables fiber-optic sensors to achieve localization and tracking directly. In order to capture more information of human subjects and scenes, a Bernoulli mixture model is proposed to model scenes. Meanwhile, compressive sensing based space encoding and decoding techniques are developed to implement scene recognition. Experimental results have demonstrated that the proposed fiber-optic sensing system and compressive sensing based encoding/decoding techniques are effective for human monitoring in terms of tracking and scene recognition.