Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A multi-view SAR image target recognition method based on deep neural network

A deep neural network and image technology, applied in the field of multi-view SAR image target recognition, can solve the problem of not being able to make full use of image correlation

Active Publication Date: 2019-01-08
BEIHANG UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing multi-view SAR image recognition methods, such as joint sparse representation, decision-level fusion, etc., cannot make full use of the correlation between images to improve the accuracy of recognition

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 multi-view SAR image target recognition method based on deep neural network
  • A multi-view SAR image target recognition method based on deep neural network
  • A multi-view SAR image target recognition method based on deep neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0122] A multi-view SAR image target recognition method based on deep neural network, specifically:

[0123] Step 1. Perform preprocessing such as size cutting and energy normalization on the input training set images and test set images.

[0124] Select three data sets of T72, BMP2, and BTR70 in the MSATR database, among which the training set is T72_132, BMP2_S71, and BTR70_C71 collected at a 17-degree viewing angle, and the test set is T72_132 collected at a 15-degree viewing angle. T72_812, T72_S7, BMP2_9563, BMP2_9566, BMP2_S71, BTR70_C71 these 7 data sets.

[0125] (1) Crop the original training image and the obtained test images of various resolutions, from 128×128 to 64×64.

[0126] (2) The energy normalization method is used to normalize the training set images and test set images. The formula is as follows

[0127]

[0128] Step 2. Construct a convolutional sparse auto-encoder (CAE for short) including a convolutional layer and a downsampling layer, and use uns...

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 multi-view SAR image target recognition method based on a deep neural network, including three steps of image preprocessing, CAE-based feature extraction, and RNN-based multi-view SAR image recognition. Firstly, preprocessing processes such as cropping and energy normalization are performed on the input graphics, and then the features of the original image are extracted through unsupervised training of CAE, and then the multi-view SAR image feature sequence is constructed by using the above features. Afterwards, supervised training of the RNN is performed with the training set feature sequence. After the training is completed, the RNN can be used to identify the feature sequence of the test set. The invention can make full use of the ability of CNN in learning and extracting image general features, and the ability of RNN to fully extract sequence context, thereby effectively improving the recognition rate of multi-view SAR image targets and having high engineering value.

Description

technical field [0001] The present invention relates to the field of radar technology, in particular to a multi-view SAR image target recognition method based on a deep neural network. Background technique [0002] As a component of SAR image interpretation system, SAR automatic target recognition has attracted extensive attention of researchers because of its significance in military and civilian fields such as disaster assessment, resource detection, and battlefield reconnaissance. SAR automatic target recognition mainly includes two parts: feature extraction and recognizer construction. For feature extraction, methods such as PCA, KPCA, and KLDA have been successfully used. For the field of target recognition, template matching method, HMM, SVM and other methods have also been tried. However, for feature extraction, the current methods mainly focus on spatial transformation processing of image features, so that different types of features have a better distinction. How...

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): G06K9/62G06N3/08
CPCG06N3/08G06F18/24323G06F18/241
Inventor 王鹏波李轩李春升门志荣
Owner BEIHANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com