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

Hyperspectral Image Compression Method Based on Distributed Compressive Sensing

A hyperspectral image and compressed sensing technology, which is applied in the field of hyperspectral image compression, can solve the problems of insufficient redundancy removal, high calculation and complexity, and serious block effects of reconstructed images, and achieve enhanced anti-error performance, Effect of protecting useful information and improving accuracy

Inactive Publication Date: 2019-06-14
NORTHEASTERN UNIV LIAONING
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the existing technology, the 3D-SPIHT compression algorithm that can support the protection of information of interest in hyperspectral images is essentially a method based on transformation. The block effect of the reconstructed image is relatively serious. In addition, the 3D-SPIHT compression algorithm fails to make full use of the strong spectral correlation of the hyperspectral image, and the removal of redundancy is not thorough enough.

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 Compression Method Based on Distributed Compressive Sensing
  • Hyperspectral Image Compression Method Based on Distributed Compressive Sensing
  • Hyperspectral Image Compression Method Based on Distributed Compressive Sensing

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0090] Such as figure 1 As shown, this embodiment provides a hyperspectral image compression method based on distributed compressed sensing, and the method includes:

[0091] Step 01: For the hyperspectral image to be processed, based on the correlation of each band in the hyperspectral image and the information entropy value of each band, divide all the bands of the hyperspectral image into a low-correlation band group and multiple high-correlation bands Group, each high correlation band group includes: a reference band and multiple non-reference bands;

[0092] For example, this step 01 may include:

[0093] 011. Obtain the information entropy values ​​of all bands in the hyperspectral image; combine the bands corresponding to the information entropy values ​​greater than the first preset entropy value into a first set s 1 ;

[0094] 012. Obtain correlation coefficients r between all adjacent bands in the hyperspectral image; determine two bands corresponding to each correlation coe...

Embodiment 2

[0130] The traditional entropy method is used to select the band with rich information. Generally, these bands are treated specially. For example, in prediction-based technologies such as differential pulse modulation (DPCM), the selected band is generally used as a reference band to predict other non-reference bands. The accuracy of the prediction will be affected by the correlation between the selected band and other bands in the group. If the band selected according to the entropy method has little correlation with the remaining bands in the group, the corresponding prediction error will be large . From the analysis of the characteristics of hyperspectral images, the correlation and entropy curve between adjacent bands of hyperspectral images have similar trends, but the correlation curve and entropy curve do not completely overlap, and even the opposite occurs at the turning point of the curve. Trend, that is, when a certain band has a large entropy value, the correlation ...

Embodiment 3

[0200] The following describes the advantages of this embodiment in combination with experimental data and experimental results:

[0201] In the experiment, two hyperspectral images of Terrain and Cuprite are selected as objects, the sparse basis is DWT, the random observation matrix is ​​part of the Hadamard matrix, and the CS reconstruction method is the Basis Pursuit (BP) method.

[0202] The result of band selection remains unchanged, and the grouping situation is shown in Table 1 and Table 2. Sampling rate SR=M / N×100%, M is the number of rows of the observation matrix, and N is the length of the original signal after thinning. Under the same conditions (Intel single-core 2.66GHz / 32-bit operating system memory 2GB), the 3D-SPIHT algorithm, the IOI-DCS algorithm (using a fixed band grouping algorithm), and the CE-DCS algorithm are compared in this embodiment.

[0203] Table 3 and Table 4 respectively show the average peak signal-to-noise ratio (Average PSNR, APSNR) (average compr...

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 provides a distributed compression sensing-based hyperspectral image compression method. The method comprises the steps of based on correlation of each band spectrum and information entropy of each band spectrum in a hyperspectral image, dividing all bands into a low correlation band group and multiple high correlation band groups, wherein each high correlation band group comprises a reference band and multiple non-reference bands; determining a region of interest and a background region in the reference band; for each non-reference band, performing differential processing on the region of interest and background region in the reference band, so as to acquire corresponding residual images; sequentially compressing residual images and low correlation band images of the region of interest and background region of each high correlation band group and each non-reference band; and sending code streams of all compression codes. According to the method, different distributed compression processing is performed on different bands and different regions, so that important information of the hyperspectral image is fully protected, and the compression rate of the hyperspectral image is improved.

Description

Technical field [0001] The invention relates to a hyperspectral image compression technology, in particular to a hyperspectral image compression method based on distributed compressed sensing. Background technique [0002] Hyperspectral image is a collection of multiple band images containing both spatial and spectral information. It has been applied to many fields, such as agriculture, military, geological prospecting, and environmental monitoring. However, with the continuous improvement of spatial resolution and spectral resolution, massive amounts of data have been brought. These massive amounts of data have brought huge challenges to the storage, transmission and application of hyperspectral images. Therefore, how to efficiently achieve hyperspectral data compression has become an urgent problem. [0003] However, among the many hyperspectral image compression algorithms proposed, most of the algorithms use the same processing for most of the bands and spatial regions in 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): H04N19/157H04N19/167H04N19/17H04N19/85
CPCH04N19/157H04N19/167H04N19/17H04N19/85
Inventor 郎俊葛锋安继成
Owner NORTHEASTERN UNIV LIAONING
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