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

Space target recognition method based on deep learning

A technology of space target and recognition method, which is applied in the field of space target recognition based on deep learning, can solve the problems of deep model overfitting and good practicability, and achieve the goal of improving recognition accuracy, good practicability and increasing scale Effect

Inactive Publication Date: 2017-11-03
NORTHWESTERN POLYTECHNICAL UNIV
View PDF9 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At the same time, in view of the fact that the spatial target data set is limited by its imaging environment and is a typical small sample problem, using data augmentation to generate virtual data can solve the problem that the depth model is easy to overfit on a small sample, and has good practicability

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
  • Space target recognition method based on deep learning
  • Space target recognition method based on deep learning
  • Space target recognition method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] The specific steps of the space target recognition method based on deep learning of the present invention are as follows:

[0020] The space object recognition problem solved by the present invention is based on the STK space object data set, which is generated by STK (System Tool Kit) satellite toolbox simulation, and degrades the original image through a series of motion blur and out-of-focus blur to simulate the real Space Imaging Environment. The data set has a total of 400 images, including four different grayscale satellite images, and each category is 100 different attitudes of each satellite.

[0021] The method of the present invention is divided into two parts, respectively constructing a 9-layer deep convolutional neural network and selecting the optimal data augmentation method.

[0022] Step 1. Construct a deep convolutional network model.

[0023] The present invention builds a 9-layer deep convolutional network on the basis of the classic LeNet-5, inclu...

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 space target recognition method based on deep learning. The objective of the invention is to solve the technical problem of poor practicability of the current space target recognition methods. The method comprises steps of firstly, constructing a 9-layer depth convolution grid model; based on the grid, finding out an optimal data augmentation method; combining data obtained through several comparatively excellent augmentation methods; and using the optimal combined data in model training and testing processes so as to finish space target recognition. According to the invention, a deep learning model automatically finds distributed characteristic representations from the data to obtain characteristics which are quite beneficial to the classification, so recognition accuracy can be greatly improved; meanwhile; based on the characteristic that a space target data set is limited by an imaging environment, which is a typical small sample problem, virtual data is generated by use of the data augmentation, so a problem that a depth model is easily over-fitted in the small samples can be solved and the method is highly practical.

Description

technical field [0001] The present invention relates to a space target recognition method, in particular to a space target recognition method based on deep learning. Background technique [0002] As a key task to ensure space security and aerospace exploration, space target recognition aims to detect and track meteorites and man-made targets distributed in near-Earth space (including space stations, aerospace vehicles, effective and invalid artificial satellites, launch vehicles, fuel tanks, etc.) and its fragments, etc.). In recent years, this task has been extensively studied and many related solutions have emerged. For example, F.Wu in the document "Research on method of space target recognition in digital image, in: Image and Signal Processing (CISP), 2012 5th International Congress on, 2012, pp.1303-1306." In the document "Research on method of space target recognition in digital image, in: Image and Signal Processing (CISP), 2012 5th International Congress on, 2012, p...

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): G06K9/00G06N3/04G06N3/08
CPCG06N3/08G06V20/13G06N3/045
Inventor 夏勇曾皓月张艳宁
Owner NORTHWESTERN POLYTECHNICAL 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