Authors: HUAZHU WU, ZENGCAI WANG, CHANGYOU WANG
Abstract: Currently, detection technology is very important for airport perimeter security. When the perimeter is invaded or destroyed, the perimeter security alarm system can promptly alert personnel. In this paper, based on analysis and comparison of several detection technologies commonly used in airport perimeter security and according to the characteristics of airport perimeters and laser detection, an airport perimeter security alarm system based on laser detection is proposed. It analyzes factors that affect the performance of a laser alarm system, divides intrusions into six categories, estimates the different alarm thresholds by testing, and judges the intrusion category according to the number of blocked laser beams and the duration of the block. In order to improve the rapid response and the robustness of the system, it uses a pattern recognition method based on radial basis function neural network technology to train and learn different samples of intrusion categories, uses the classifier to determine the cluster center of each sample node, and uses the weights of the center variance and the weights of hidden nodes to establish the alarm curve of each intrusion category. Meanwhile, when the feature value of an intrusion sample is distributed along the classification boundary of two categories, it uses Bayes' theorem to calculate the probability of an invasive category to reduce the false alarm risk. According to sample testing, the above recognition method of airport intrusions based on laser detection technology effectively improves the intrusion detection probability.
Keywords: Pattern recognition, neural network, Bayesian theory, laser detection
Full Text: PDF