Remote sensing image target sample generation method for deep learning

A technology of remote sensing images and deep learning, applied in the field of deep learning and target recognition, remote sensing, can solve the problems of unfavorable detection and recognition accuracy, loss of detail features, etc., achieve good performance of remote sensing image target detection and recognition, improve performance, increase samples The effect of diversity

Inactive Publication Date: 2019-10-15
NO 54 INST OF CHINA ELECTRONICS SCI & TECH GRP
View PDF9 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method can obtain more abstract features of the target, but some of its detailed features are lost, which is not conducive to improving the accuracy of detection and 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
  • Remote sensing image target sample generation method for deep learning
  • Remote sensing image target sample generation method for deep learning
  • Remote sensing image target sample generation method for deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0040] A deep learning-oriented remote sensing image target sample generation method, which includes the following steps:

[0041] (1) Extract sample slices through wide-range remote sensing images, and obtain measured remote sensing image slices containing targets;

[0042] (2) Mark the position and category of the target in the measured remote sensing image slice;

[0043] (3) Generate remote sensing image target sample sets according to the marked remote sensing image slices;

[0044] (4) Evaluate the remote sensing image target sample set generated in step (3) based on information entropy;

[0045] (5) According to the evaluation result of step (4), the remote sensing image target sample set is selected. If the remote sensing image target sample set does not meet the requirements, delete the remote sensing image target sample set, and ret...

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 remote sensing image target sample generation method for deep learning, and belongs to the technical field of remote sensing. The method comprises the steps of extracting a sample slice, marking the position and the category of a target in an actually measured remote sensing image slice, generating a remote sensing image target sample set, evaluating the remote sensing image target sample set based on information entropy, selecting and rejecting the remote sensing image target sample set according to an evaluation result and the like. The method has the characteristics of high efficiency, strong sample descriptive property, good robustness and the like, and is suitable for the application fields of target detection, target positioning, sample preparation preprocessing of target identification and the like of visible light remote sensing images.

Description

technical field [0001] The invention relates to the technical fields of remote sensing, deep learning, and target recognition, and in particular to a deep learning-oriented remote sensing image target sample generation method. Background technique [0002] In the field of remote sensing, there are mainly the following methods for training sample augmentation, but they all have some defects in performance and efficiency: [0003] (1) The method of sliding window extraction directly obtains the corresponding target image data according to the size and amplitude of the sliding window. This method is the simplest, but the target shape does not change, it is less helpful for subsequent feature learning, and the background information is missing. [0004] (2) Perform mirror transformation directly on the image. This method retains all the information of the target and the background, but the target changes less and is less helpful for subsequent feature learning. [0005] (3) R...

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): G06K9/00G06K9/62G06K9/32
CPCG06V20/13G06V10/242G06F18/214
Inventor 王港王敏高峰陈金勇
Owner NO 54 INST OF CHINA ELECTRONICS SCI & TECH GRP
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