论文部分内容阅读
在宽域全光纤安防系统中,在不影响灵敏度的情况下区分入侵和正常事件是一个关键性的系统性能指标。由于入侵和正常事件以及各个入侵事件的光纤信号在某些情况下极为相似,因此需要对这些信号的特征进行仔细的筛选和识别。本文由此出发,提出了一套光纤信号特征提取方案,即使用小波降噪手段对光纤信号进行去噪;根据信号与噪声的能量在时域分布的不同,提出了一种实用的光纤信号的预分割方法;提取信号在小波空间的能量分布特征,形成特征向量;使用向量机分类器对光纤信号分类。
In wide-area all-fiber security systems, distinguishing between intrusions and normal events without compromising sensitivity is a key system performance metric. Since intrusion and normal events and fiber-optic signals at various intrusion events are very similar in some cases, the characteristics of these signals need to be carefully screened and identified. In this paper, a set of optical signal feature extraction scheme is proposed, that is, the wavelet denoising is used to denoise the optical fiber signal. According to the difference of signal energy and noise energy in the time domain, a practical optical fiber signal Pre-segmentation method; extract the energy distribution characteristics of the signal in the wavelet space to form eigenvectors; classify the optical fiber signal using a vector machine classifier.