基于K-means聚类的Morse码自动译码研究

Study of Automatic Decoding for Morse Code Based on K-means Clustering

针对短波Morse电报自动译码准确率低的问题,提出了一种针对信号时频图,采用机器学习理论的Morse信号自动译码技术,实现了对机械或手工Morse码的自动译码。通过对短时傅里叶变换生成的Morse信号时频图像,进行自适应图像增强处理,提取Morse信号;采用基于K-means聚类的非监督学习方法自动分类Morse点划、间隔等信息,实现Morse码自动译码;为提高译码的正确率,设计了译码纠错方法。仿真试验和工程实践均验证了算法的有效性。

To solve the technical problems of low accuracy of shortwave Morse telegraph automatic decoding,an automatic manual or mechanical Morse decoding method for time-frequency spectrum of signal is put forward based on machine learning technology.Time-frequency spectrum is generated based on STFT,which is used for extraction of Morse signal based on adaptive image enhancement. Unsupervised learning method based on K-means clustering has been introduced to identify the dots,dashes and intervals between them. Error correction algorithm is put forward to improve the accuracy of decoding. Engineering practice and simulation on Morse signal demonstrate the effectiveness of this algorithm.