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

Hyperspectral Data Classification Method Based on Convolutional Neural Networks to Convert Space-Spectrum Joint Data to Waveform Image

A convolutional neural network and joint data technology, applied in neural learning methods, biological neural network models, instruments, etc., can solve the problems of long classification time, low data classification accuracy, and large amount of calculation, and achieve high classification accuracy. , enrich the data fluctuation characteristics, improve the effect of robustness

Active Publication Date: 2019-07-02
HARBIN INST OF TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to solve the problems of low accuracy rate of existing data classification, long classification time, and large amount of calculation, and proposes a hyperspectral data classification method based on convolutional neural network-based space-spectrum joint data to waveform diagram

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
  • Hyperspectral Data Classification Method Based on Convolutional Neural Networks to Convert Space-Spectrum Joint Data to Waveform Image
  • Hyperspectral Data Classification Method Based on Convolutional Neural Networks to Convert Space-Spectrum Joint Data to Waveform Image
  • Hyperspectral Data Classification Method Based on Convolutional Neural Networks to Convert Space-Spectrum Joint Data to Waveform Image

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0027] Specific implementation mode 1: The hyperspectral data classification method based on the convolutional neural network-based space-spectrum joint data conversion waveform diagram in this embodiment is specifically carried out in accordance with the following steps:

[0028] Step 1. Normalize the hyperspectral raw data layer by layer to obtain normalized hyperspectral data;

[0029] Process the normalized hyperspectral data to obtain hyperspectral space-spectrum joint data information;

[0030] Step 2, converting hyperspectral space-spectrum joint data information into two-dimensional waveform image data.

specific Embodiment approach 2

[0031] Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that in the step 1, the hyperspectral raw data is normalized layer by layer to obtain normalized hyperspectral data; Hyperspectral data is processed to obtain hyperspectral space-spectrum joint data information; the specific process is:

[0032] Such as Figure 7 As shown, when the pixel to be processed in the normalized hyperspectral data is not located at the edge of the hyperspectral remote sensing image, the field splicing method is used to combine the pixel to be processed in the hyperspectral data with the surrounding pixel by 5*5 The spectral information splicing of the square neighborhood space of large and small sizes constitutes the hyperspectral space-spectrum joint data information;

[0033] When the pixel to be processed in the normalized hyperspectral data is located at the edge of the hyperspectral remote sensing image, the spectral information of a 5*5 square neig...

specific Embodiment approach 3

[0036] Specific embodiment three: the difference between this embodiment and specific embodiment one or two is: in the first step, hyperspectral raw data is normalized layer by layer to obtain normalized hyperspectral data; the specific process is:

[0037]

[0038] In the formula, is the normalized hyperspectral data; is the hyperspectral raw data at the position (i, j) of the kth layer; W is the width of the hyperspectral raw data; L is the length of the hyperspectral raw data; H is the depth of the hyperspectral raw data; is a positive integer; 1≤i≤W, 1≤j≤L, 1≤k≤H.

[0039] The present invention uses layer-by-layer internal normalization. If normalization is not used, part of the spectrum information will be ignored. The specific situation is as follows Figure 8 Shown:

[0040] Such as Figure 8 As shown in , taking a matrix with four spectral segments as an example, the data values ​​of the first spectral segment are generally too large, while the data values ​​o...

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 relates to a hyperspectral data classification method for converting space-spectrum joint data into wave form images based on a convolutional neural network. The purpose of the present invention is to solve the problems of low classification accuracy rate, long classification time and large calculation amount of existing data. The hyperspectral data classification method based on the convolutional neural network to transform the space-spectrum joint data into a waveform image is specifically carried out in accordance with the following steps: Step 1. Normalize the hyperspectral raw data layer by layer to obtain the normalized hyperspectral data. Data; processing the normalized hyperspectral data to obtain hyperspectral space-spectrum joint data information; Step 2, converting the hyperspectral space-spectrum joint data information into two-dimensional waveform image data. The invention is used in the field of hyperspectral data classification.

Description

technical field [0001] The invention relates to a hyperspectral data classification method for converting space-spectrum joint data into waveform images. Background technique [0002] Hyperspectral data classification is an application of hyperspectral remote sensing. It uses the characteristics that all objects have spectral characteristics, and objects in the same spectral region respond differently, and the same object responds differently to different spectra. classification of remote sensing data. Spectral images with a spectral resolution within the order of magnitude of 10l are called hyperspectral images (Hyperspectral Image). The space-spectrum joint classification method of hyperspectral data is a method to classify hyperspectral data by using the spatial information of hyperspectral data. In real life, the spatial distribution of ground objects has a very high regularity, often accompanied by a certain spatial structure. When performing data classification, the...

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
CPCG06N3/08G06F18/2411
Inventor 林连雷宋欣益魏长安
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