Unlock instant, AI-driven research and patent intelligence for your innovation.

A Method for Matching and Recognition of Individual Targets in Water Based on Convolutional Neural Networks

A convolutional neural network and target matching technology, which is applied in the underwater acoustic target classification and recognition technology and the field of artificial intelligence, can solve problems such as poor generalization ability, difficulty in extracting acoustic signal features, weak environmental adaptability, etc., to achieve effective recognition, strong The effect of nonlinear data processing capabilities, high accuracy and robustness

Active Publication Date: 2021-08-17
THE 715TH RES INST OF CHINA SHIPBUILDING IND CORP
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Traditional feature extraction methods start from physical mechanisms and phenomena, and extract physical features through signal processing and transformation. Therefore, it is difficult to extract acoustic signal features that can describe clear individual differences, and there are deficiencies such as weak environmental adaptability and poor generalization ability. Actual use requirements and operational requirements

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
  • A Method for Matching and Recognition of Individual Targets in Water Based on Convolutional Neural Networks
  • A Method for Matching and Recognition of Individual Targets in Water Based on Convolutional Neural Networks
  • A Method for Matching and Recognition of Individual Targets in Water Based on Convolutional Neural Networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0040] EXAMPLES: As shown in the drawings, this individual target matching recognition method based on convolutional neural networks, mainly includes the following steps:

[0041] 1) Based on the TENSORFLOW framework constructing a convolutional neural network model for feature extraction, it is mainly constructed of three basic modules, constructing the entire convolutional neural network, set the Dropout coefficient of 0.25 in each convolution operation, activation function Using the RELU function, use the Triplet LOSS method to build a loss function, set the training parameters such as optimizer, learning rate, and training at the time of iterative training;

[0042] The main steps in which 3 basic modules are built are as follows:

[0043] Step 1: Build the basic module 1, add 4 parallel branches after the data input layer, the branch 1 is the direct branch, does not add any operation, the branch 2 includes three convolution layers, the convolution layer 1 parameter is (1 × 1,...

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 discloses a method for matching and identifying underwater individual targets based on a convolutional neural network. Firstly, a convolutional neural network model is constructed for feature extraction of time-frequency image features of radiation noise of underwater targets, and secondly, the tagged target noise is generated through S-transform preprocessing. The time-frequency image data is used to train and improve the convolutional neural network model, and then the time-frequency image reference data with labeled target noise is generated based on the S-transform, and the individual target numerical difference feature template is obtained through convolutional neural network processing. The network processes the unknown target data to obtain numerical features, and uses the template matching method to identify individual targets. Compared with traditional feature extraction and recognition methods based on physical mechanisms, the present invention has stronger nonlinear data processing capability and can more effectively identify individual targets in water.

Description

Technical field [0001] The present invention relates to the field of water sound target classification identification techniques and artificial intelligence technologies, and in particular to an individual target matching recognition method based on convolutional neural network. Background technique [0002] The target identification technology in the water is the information processing technology of target radiation noise and other sensor information discriminating target types, which can be used to provide target feature information, discrimination target type, and is an important basis for the comprehensive decision-making. Traditional target recognition typically divides the goal of water into several categories, such as water surface ships, submarines, merchant ships, torpedoes, and more focuses on target classification, and individual target identification requires further identification of target specific models and jewels based on the type of resolution target. , A higher...

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 Patents(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62
CPCG06V20/00G06V10/454G06F18/21355G06F18/2453G06F18/214
Inventor 陈越超尚金涛
Owner THE 715TH RES INST OF CHINA SHIPBUILDING IND CORP
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com