基于改进K均值聚类算法的摩斯报抄报研究

Research on Morse Telegraph Decoding based on Improved K-Means Clustering Algorithm

摩斯电报在紧急情况下有无可替代的重要作用。如今的摩斯报自动抄报方法尚不成熟,而手工抄报方法会对抄报人的身体造成一定负担。因此,针对摩斯报自动抄报存在的问题,提出一种基于改进聚类算法的摩斯报抄报方法。通过时频分析方法和基于阈值的图像分割方法得到摩斯信号的时频二值图,通过机器学习中的聚类算法将时频二值图识别为点、划与间隔的组合。根据摩斯码的特性改进K均值聚类算法,使其更适用于摩斯报的自动抄报,在保持原有算法高准确率的同时,大幅提升了运算速度。

Morse telegraph plays an irreplaceable role in emergency,but the Morse telegraph automatic decoding method is not mature,and the manual method will cause a certain burden on the body of the person who copy the telegram.To solve the problem,a Morse telegraph decoding algorithm based on improved clustering is proposed.The time-frequency binary graph of Morse signal is obtained by time-frequency analysis method and image segmentation method based on threshold.The time-frequency binary graph is identified as the combination of point,stroke and interval by clustering algorithm in machine learning.According to the characteristics of Morse code,the K-means clustering algorithm is improved to make it more suitable for Morse code automatic reading.While maintaining the high accuracy of the original algorithm,the operation speed is greatly improved.