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

Hyperspectral Image Lossless Compression Method Based on Deep Learning

A hyperspectral image, lossless compression technology, applied in the field of image processing, can solve the problems of model generalization ability and prediction accuracy need to be improved, information utilization is not very sufficient, etc., to increase generalization ability, improve compression bit rate, improve The effect of compression precision

Active Publication Date: 2021-12-17
XIDIAN UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] DPCM realizes the lossless compression of hyperspectral images, but its method does not fully utilize the information of the spectral bands used for prediction, and the generalization ability and prediction accuracy of the model need to be improved

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 Image Lossless Compression Method Based on Deep Learning
  • Hyperspectral Image Lossless Compression Method Based on Deep Learning
  • Hyperspectral Image Lossless Compression Method Based on Deep Learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0024] At present, hyperspectral remote sensing technology is at the forefront of remote sensing technology, and a lot of manpower and material resources have been invested at home and abroad to study it, but there are still huge difficulties in the real-time transmission of data acquired by hyperspectral imagers. In response to this problem, the present invention conducts research from the perspective of image lossless compression, focuses on more efficient lossless compression methods, and proposes a hyperspectral image lossless compression method based on deep learning, see figure 1 , including the following steps:

[0025] (1) Clustering all spectral lines in the original hyperspectral image: input the hyperspectral image to be compressed, which has M bands. The standard deviation of hyperspectral image data bands is large and the amount of information is rich; while the correlation coefficient between bands is small and the redundancy of band information is small. In this...

Embodiment 2

[0032] The hyperspectral image lossless compression method based on deep learning is the same as embodiment 1, and the spectral line after clustering described in step (2) is predicted using a deep recurrent neural network, including the following steps:

[0033] (2a) Intra-spectrum prediction is used for the first band, that is, adjacent pixel prediction is used for intra-band prediction, that is, the pixel value of the first band of the first spectral line is directly encoded and transmitted as the residual, and the pixel value of the second spectral line is directly encoded and transmitted. A band is predicted from the first spectral line, and in this way the first bands of all spectral lines are obtained. The adjacent pixel prediction used in this example is to use the core algorithm LOCO-I of the JPEG-LS compression standard to perform intra-band prediction. The formula is

[0034]

[0035] (2b) From the second wave band, the recurrent neural network is used to carry o...

Embodiment 3

[0041] The hyperspectral image lossless compression method based on deep learning is the same as embodiment 1-2, the network structure of the recurrent neural network described in step (2b), and the structure of the neural network is shown in the appendix image 3 As shown, it includes the following parts in order from input to output: 1 input layer, 3 hidden layers, 1 fully connected layer, and an output layer, using softmax for classification. In this example, an expanded network is used. The input layer has 20 units, and each hidden layer has 40 units. The fully connected layer after the hidden layer fuses the features extracted from the hidden layer to obtain a unit in the output layer.

[0042] Compared with the traditional method, the present invention has more efficient learning efficiency. The hidden layer uses an expanded network, that is, the number of units used in the hidden layer is input to the layer, which increases the ability of the model to extract features and ...

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 hyperspectral image lossless compression method based on deep learning, which solves the problems of insufficient utilization of spectral information and low model generalization ability in traditional methods. The implementation steps include: applying the cyclic neural network in deep learning to establish a prediction model, performing prediction training for each pixel in the hyperspectral image, generating a prediction image and a prediction network; making a difference between the hyperspectral image and the prediction image, and generating a residual image; Perform arithmetic coding on the residual image to generate a code stream file; decode the code stream file to obtain a decoded image; use the trained network for prediction to obtain a predicted image; add the predicted image to the decoded image to obtain the original hyperspectral image . The invention combines deep learning with traditional methods, adopts units with memory structure in the network, and fully utilizes spectral information through a large number of trainings, increases the generalization ability of the model, and improves the compression efficiency. It is used in the field of hyperspectral image compression.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to image lossless compression, in particular to a deep learning-based lossless compression method for hyperspectral images, which is used for hyperspectral image compression. Background technique [0002] With the rapid development of hyperspectral remote sensing technology, the spatial resolution and inter-spectral resolution of hyperspectral images are also continuously improving, resulting in the continuous increase of the scale of hyperspectral data. Usually imaging spectrometers use airborne or spaceborne methods to obtain impact data, so for satellite data links, the contradiction between massive hyperspectral image data acquisition and transmission is increasingly intensified, thus restricting the development of hyperspectral remote sensing technology. Therefore, for the long-term development of hyperspectral remote sensing technology, it is particularly important to s...

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): H04N19/593
CPCH04N19/593G06N3/08G06F18/214
Inventor 吴家骥赵世慧屈檀李皎皎
Owner XIDIAN 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