Pattern Recognition and Mixed Programming in All Fiber Security Monitoring System
Xiaohan Sun · Microcomputer Information · 2011
This paper presents an all fiber security monitoring system and pattern recognition algorithm on fiber disturbance signal.Wavelet packet is a method of analyzing and reconstructing signal,we apply it to extract feature vector.Classification is realized by RBF neural network,whose speed and precision are both quit good.Pattern recognition algorithm is developed with MATALB and integrated into the application developed with VC through two mixed programming methods:MATLAB engine,and component objective model(COM).We read and write network data through text file to overcome the limitation that MATLAB can not compile functions in neural network toolbox.Operation complexity,running time and memory consumption of these methods are compared and component objective model is the final choose.