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

High-spatial-resolution remote sensing image scene classification method based on target enhancement

A high spatial resolution, remote sensing image technology, applied in the field of high spatial resolution remote sensing image scene classification based on target enhancement, can solve the problems of low classification accuracy of remote sensing image scene classification, low efficiency of manual classification methods, complex spatial distribution, etc. Achieve the effects of improving classification efficiency, accelerating convergence speed, and improving learning efficiency

Active Publication Date: 2020-09-22
HARBIN INST OF TECH
View PDF4 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem of low classification accuracy of remote sensing image scene classification and low efficiency of manual classification method due to various forms and complex spatial distribution of ground objects in high spatial resolution remote sensing images, and proposes A Scene Classification Method for High Spatial Resolution Remote Sensing Images Based on Target Enhancement

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
  • High-spatial-resolution remote sensing image scene classification method based on target enhancement
  • High-spatial-resolution remote sensing image scene classification method based on target enhancement
  • High-spatial-resolution remote sensing image scene classification method based on target enhancement

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0033] Specific implementation mode 1: Combination figure 1 This embodiment will be described. The method for classifying high spatial resolution remote sensing image scenes based on target enhancement described in this embodiment specifically includes the following steps:

[0034] Step 1: Collect a set of high spatial resolution remote sensing images X 1 (Generally, the resolution above 5m can be regarded as a high spatial resolution image), the high spatial resolution remote sensing image X 1 The label vector of the scene category contained in is Y;

[0035] Step two, to the collected high spatial resolution remote sensing image X 1 Gaussian filtering is performed on each image in respectively to obtain the Gaussian filtered image corresponding to each image;

[0036] Then convert each Gaussian filtered image into a Lab image (color space image). For any Lab image, calculate the average value of all pixels in the Lab image on each channel, and convert all pixels in the Lab image ...

specific Embodiment approach 2

[0044] Specific implementation manner two: combination Figure 2a with Figure 2b This embodiment will be described. The difference between this embodiment and the first embodiment is that the specific process of the second step is:

[0045] Step Two: Respectively perform high spatial resolution remote sensing image X 1 Perform Gaussian filtering on each image in, to obtain an image after Gaussian filtering;

[0046] The high-frequency information in the image is removed from the image after Gaussian filtering, so that the image has more low-dimensional spatial information, that is, the image becomes smoother.

[0047] The image processed by Gaussian filtering is an RGB image, and each RGB image is converted into a corresponding Lab image. The conversion formula is as follows:

[0048]

[0049]

[0050]

[0051] In the formula, R, G, B are the elements of the RGB image on the three bands, L, a, b are the elements of the Lab image on the three channels, X, Y, Z, L', M'and S ′ Are inte...

specific Embodiment approach 3

[0066] Specific implementation mode three: combination Figure 3a This embodiment will be described. The difference between this embodiment and the second embodiment is that the specific process of the third step is:

[0067] Use the attention mask matrix as the attention weight, through the feature information block F and the initialized attention mask matrix a 1 Calculate the initial weight value x in the input long and short-term memory network (LSTM) 1 And an enhanced feature information block F 1 ;

[0068] Where the initialized attention mask matrix a 1 The value of is randomly generated;

[0069] The specific calculation formula is as follows:

[0070] a 1 ={a 1,1 ,a 1,2 ,...,A 1,P×P }

[0071]

[0072] f 1,j =a 1,j ×f j ,a 1,j A 1 ,f j ∈F,f 1,j ∈ F 1 ,j∈1,2,…,P×P

[0073] Where a 1 Is the initial attention mask matrix, a 1,j Is the jth element in the initial attention mask matrix, f 1,j Is an enhanced feature information block F 1 The jth element in x 1 It is the initial weight...

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 high-spatial-resolution remote sensing image scene classification method based on target enhancement, and belongs to the technical field of remote sensing image scene classification. According to the method, the problems of low image scene classification precision and low efficiency of a manual classification method due to various ground object forms and complex spatial distribution in the high-spatial-resolution remote sensing image are solved. According to the method, saliency enhancement processing is carried out on the high-resolution remote sensing image by usinga saliency mechanism in computer vision. A repeated attention structure is provided, and an effective high-spatial-resolution remote sensing image scene classification method based on target enhancement is constructed on the basis of the repeated attention structure. The method is inspired by an attention mechanism of a human vision system, image salient features are enhanced in an image featureiteration mode, and then continuous learning is performed to focus on an image key region, so that the classification precision can be effectively improved, the convergence speed can be increased, andthe learning efficiency can be improved. The method can be applied to remote sensing image scene classification.

Description

Technical field [0001] The invention belongs to the technical field of remote sensing image scene classification, and in particular relates to a high spatial resolution remote sensing image scene classification method based on target enhancement. Background technique [0002] At this stage, with the rapid development of remote sensing detection methods, a series of commercial high-resolution remote sensing satellites such as Quickbird, Worldview series, GeoEye series, domestic GF-1, etc. have been launched one after another, making it easier to obtain high spatial resolution remote sensing images And the application of remote sensing images with high spatial resolution is becoming more and more popular. As an important part of remote sensing technology, remote sensing image scene classification is widely used in military and civilian fields such as homeland security monitoring, land cover / land use classification, urban planning, and environmental monitoring. With the continuous ...

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
CPCG06N3/049G06N3/08G06V20/13G06V20/41G06N3/044G06N3/045G06F18/2415
Inventor 谷延锋白洋高国明
Owner HARBIN INST OF TECH
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