MorseNet: A Unified Neural Network for Morse Detection and Recognition in Spectrogram

Abstract

Short-wave radio is an indispensable long-distance means of communication, among which Morse signals, which rely on simplicity and efficiency, plays an import role in military and civilian applications. Automatic Morse detection and recognition have been researched for several years, but some thorny problems in actual communication always restrict the performance of methods. In this paper, by introducing deep learning technology, we propose a network named MorseNet that can simultaneously locate and decode Morse signals in the spectrogram. MorseNet uses shared convolutions to extract shared features for both the detection and recognition branches. The detection branch regresses bounding boxes based on signal centerlines, and the recognition branch decodes Morse fragments cropped from feature maps by a convolutional recurrent neural network (CRNN). The losses of two branches are combined to implement the end-to-end training. Experimental results on four "simulated Morse + real background" datasets demonstrate that the proposed method achieves state-of-the-art performance in both detection and recognition, and it effectively improves four problems that have long been troublesome in accomplishing the tasks. Furthermore, the joint training strategy and architecture give MorseNet advantages over its twostage deployment in terms of accuracy, speed, and model size.

Morse code detection and recognition algorithm based on YOLO-SVTR

“…However, the processing speed and decoding accuracy of this method still need to be improved. Li et al proposed an end-to-end Morse code detection and recognition network, MorseNet 9 , based on deep learning, which maintains good robustness in low signalto-noise ratios. This article also proposes the YOLO-SVTR algorithm based on deep learning technology.…”

Detection and recognition based on machine learning and deep learning for Morse signal in wide-band wireless spectrum

“…Two of them extract the signal fragments in the time-frequency spectrum through energy sorting and then design a machine learning classifier [15] or a deep neural network classifier [16] to detect Morse signals. Another method is to directly perform target detection based on deep learning on the time-frequency spectrum of audio data [17]. The channel environment of several methods is broadband, and the signal interval of radio stations is far away.…”

“…In addition, in previous studies, it was assumed at the beginning of recognition that there was only the target Morse signal in the time-frequency spectrum [20][21]25], or the Morse signal time-frequency spectrum segment detected in the detection phase was directly used as the input of the recognition phase [17,22], and there was no processing between detection and recognition. The recognition and detection have high requirements for time resolution and frequency resolution, respectively [22].…”