Human Detection Based on Radar Sensor Network in Natural Disaster

Abstract

In recent years, natural disasters, such as earthquakes, landslides and others, have caused significant damage to people’s lives and property. Victims are often trapped in collapsed buildings. Thus the development and understanding of modern techniques for disaster relief are of immense current interest and need. As a significant advancement in wireless communication, the emerging UWB Radar Technology is a key technology that UWB is applied in object identification, which is characterized by high resolution, good anti-interference ability and strong penetrability and so on, has been widely used in various fields, including natural disaster detection, through-wall radar imaging, ground penetrating radar technology, medical imaging, target ranging and personnel positioning, disaster relief and so on. In this chapter, the author will describe some algorithms for human detection based on UWB radar sensor network in natural disaster. Firstly, we study the fuzzy pattern recognition and genetic algorithm which is used to identify the multi-status human being after the brick wall. The main characteristic parameters are selected and extracted from the received signal, and each feature parameters corresponding to a sub membership function. Through the genetic algorithm to optimize the sub membership function for constructing the membership function set. According to fuzzy pattern recognition principle of maximum degree of membership function to establish target prediction function, and used MATLAB to carry on the simulation for it. Secondly, we study the stacked denoising autoencoder algorithm in deep learning to study the through wall human target recognition under imbalanced samples of single sensor and multi-sensor data respectively. The experimental results show that the stacked denoising autoencoder algorithm in deep learning adopted herein allows more effective classification and identification of through wall human targets under imbalanced sample conditions than other algorithms, and that the identification effect with multiple sensors under a certain imbalance rate is better than that with a single sensor.