Morse code automatic identification method based on Bi-LSTM neural network

A neural network and automatic identification technology, applied in the field of communication signal processing, can solve the problems of short working time, unfavorable copying, and high labor intensity of the operator, and achieve the effect of improving processing efficiency, increasing depth, and accurate processing results.

Pending Publication Date: 2020-04-28
长沙深之瞳信息科技有限公司
View PDF7 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

It can be seen from this that there are three major defects in the manual copying work method: 1. The operator's working hours are short. Generally, the manual operator has to rest after working for two hours, and cannot perform continuous copying work for a long time; 2. The operator The labor intensity is high, and the telegraph operator must have super brain and manual skills, receive and grasp the sound signal very accurately, and correct the few errors in the process of sending the machine to get a text file as the result of the message. ; 3. The operator training period is long, which is not conducive to rapid and large-scale replication
[0004] In recent years, with the popularity and popularization of artificial intelligence and deep learning, a large number of tasks that require manual repetitive labor have been replaced by artificial intelligence and deep learning technology, but the recognition of Morse code still relies on experienced operators To complete, there is no effective technical means that can use artificial intelligence and deep learning technology to complete the automatic recognition of Morse code

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Morse code automatic identification method based on Bi-LSTM neural network
  • Morse code automatic identification method based on Bi-LSTM neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0046] The present embodiment provides the automatic recognition method of Morse code based on Bi-LSTM neural network, such as Figure 1 to Figure 2 shown, including the following steps:

[0047] S1. Construct a convolutional neural network and a Bi-LSTM neural network, and combine the Bi-LSTM neural network and the convolutional neural network for sequence modeling to generate a multi-mode LSTM model;

[0048] S2. The multi-mode LSTM model is trained by means of joint training, and the parameters of the Bi-LSTM neural network and the convolutional neural network are jointly optimized;

[0049] S3. Acquire the Morse code audio signal, and preprocess the Morse code audio signal to obtain the preprocessed audio signal;

[0050] S4. Analyzing and converting the preprocessed audio signal to generate a spectrum image of the audio signal;

[0051] S5. Input the spectrum image into the multi-mode LSTM model, and output the probability vector result;

[0052] S6. Using the result o...

Embodiment 2

[0063] As an optimization to the above-mentioned embodiment, in step S1, the constructed Bi-LSTM neural network is a bidirectional multilayer Bi-LSTM neural network, and the output data of each layer in the bidirectional multilayer Bi-LSTM neural network is used as the next layer input data.

[0064] The Bi-LSTM neural network can be expressed as:

[0065] S t =f(UX t +WS t-1 )

[0066] S t ’=f(U’X t +W'S t+1 ')

[0067] OT=g(VS t +V'S t ')

[0068] Among them, S t Indicates the hidden layer state value of t spectral point, S t 'Represents the reverse hidden layer state value of t spectral point, OT represents the value of the output layer of t spectral point, S t-1 Indicates the hidden layer state value of the t-1 spectral point, S t+1 'Represents the reverse hidden layer state value of t+1 spectral point, g, f represent different activation functions, X t Represents the input vector, U represents the weight matrix from the input layer to the hidden layer, U' r...

Embodiment 3

[0071] As an optimization of the above-mentioned embodiment, in step S3, the process of preprocessing the Morse code audio signal includes volume normalization processing, signal amplification processing, high-pass filtering processing and signal noise reduction processing, and the volume normalization processing makes the volume of the audio clip Regional average, high-pass filter processing to filter out low-frequency environmental noise with a frequency lower than 300Hz.

[0072]In step S4, when converting the preprocessed audio signal, Fourier transform is used to convert the preprocessed audio signal into a spectrum image, and the sampling frequency is 8000 Hz. Before performing the Fourier transform process, first enhance the strength of key frequency signals in the audio signal (linear enhancement, coefficient 1.2), and the key frequency range is 1000Hz±100Hz.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to the technical field of communication signal processing, in particular to a Morse code automatic identification method based on a Bi-LSTM neural network. The method comprises the following steps: S1, constructing a convolutional neural network and a Bi-LSTM neural network, and performing sequence modeling in combination with the Bi-LSTM neural network and the convolutionalneural network to generate a multi-mode LSTM model; S2, training the multi-mode LSTM model in a joint training mode, and performing joint optimization on parameters of the Bi-LSTM neural network and the convolutional neural network; S3, acquiring a Morse code audio signal, and preprocessing the Morse code audio signal to obtain a preprocessed audio signal; S4, analyzing and converting the preprocessed audio signal to generate a frequency spectrum image of the audio signal; S5, inputting the frequency spectrum image into a multi-mode LSTM model, and outputting a probability vector result; and S6, judging the content of the Morse code by utilizing a probability vector result. According to the method, the deep neural network model based on Bi-LSTM can be utilized to efficiently and accuratelycomplete automatic identification of the Morse code.

Description

technical field [0001] The invention relates to the technical field of communication signal processing, in particular to an automatic recognition method of Morse code based on a Bi-LSTM neural network. Background technique [0002] Telegraph communication in the world generally adopts the Morse code method, and manual copying is used in the process of copying and receiving messages. There is no machine or equipment that can replace manual copying and receiving of messages. As an international telegraph communication symbol, Morse code uses various combinations of signals of different lengths to represent certain letters, numbers and punctuation marks. When writing, a short signal is generally represented by a dot ".", and a long signal is represented by a dash "-". In order to distinguish dots and dashes and to distinguish each character, there are strict regulations on the length of dots and dashes and various intervals: 1 dash is equal to the length of 3 dots without inte...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/044G06N3/045G06F2218/08G06F2218/12G06F18/2411G06F18/2415
Inventor 曾英夫
Owner 长沙深之瞳信息科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap