【摘要】：針對現(xiàn)有算法沒有涉及的寬帶環(huán)境下多路摩爾斯信號自動檢測問題,提出一種基于寬帶時頻圖和集成學習分類器的算法.首先,通過提出一種基于寬帶時頻圖的信號快速窄帶濾波方法,實現(xiàn)噪聲背景中各類型信號窄帶時頻圖的快速獲取;然后,為從上述窄帶時頻圖中識別出多路摩爾斯信號,提出3個新特征與局部二值模式特征構成特征向量;最后,采用集成學習算法設計分類器實現(xiàn)摩爾斯信號的自動檢測.與現(xiàn)有算法對比實驗結果表明:針對多組實際數據,該算法的正確率均可達95%以上,同時誤檢率低于10%,具有良好的魯棒性和應用價值.
[Abstract]:In order to solve the problem of automatic detection of multi-channel Morse signals in broadband environment, an algorithm based on broadband time-frequency graph and integrated learning classifier is proposed. Firstly, a fast narrowband filtering method based on broadband time-frequency graph is proposed to obtain the narrowband time-frequency graph of each type of signal in noise background. Then, in order to recognize the multi-channel Morse signal from the narrowband time-frequency graph, three new features and local binary pattern features are proposed to form the feature vector. Finally, the integrated learning algorithm is used to design a classifier to realize the automatic detection of Morse signal. Compared with the existing algorithms, the experimental results show that the accuracy of the algorithm can reach more than 95% and the error detection rate is less than 10% for multiple groups of actual data, so it has good robustness and application value.
【作者單位】： 先進信息網絡北京實驗室;北京工業(yè)大學信息學部;
【基金】：國家自然科學基金資助項目(61672064) 北京市自然科學基金資助項目(KZ201610005007)
【分類號】：TN925;TP391.41
，

本文編號：2521841