Hyperspectral Image Compression Based on DLWT and
PCA
Qiuyan Shi
Xi’an Jiaotong University
No. 28, Xianning West Road,
Xi’an , P.R. China
+86 29 82668772, 710049
qq1249738951@stu.xjtu.edu.cn
Xingsong Hou
Xi’an Jiaotong University
No. 28 , Xianning West Road,
Xi’an , P.R. China
Xueming Qian
Xi’an Jiaotong University
No. 28, Xianning West Road,
Xi’an , P.R.China
ABSTRACT
Each band, which is the image of the same object on different
frequency bands for hyperspectral image, has not only the
correlation in space, but also a strong correlation between
spectrum. The hyperspectral image compression algorithms need
to consider how to make use of the correlation of both space
and spectrum. In this paper, we first use principal component
analysis (PCA) to remove the spectral correlation. Then a
directional lifting wavelet transform(DLWT) is used to remove
the spatial correlation. The experimental results show that the
proposed image compression scheme achieves higher
performances when compared with DWT based Consultative
Committee for Space Data Systems(CCSDS).
Categories and Subject Descriptors
I.4.2[Image Processing and Computer Vision]:
Compression(Coding) –Approximate methods.
I.4.5 [Image Processing and Computer Vision]:Reconstruction–
Transform methods.
General Terms
Algorithms
Keywords
Hyperspectral image, CCSDS, DLWT, compression, PCA
1. INTRODUCTION
So far, countries around the world are racing to develop the
hyperspectral imaging technology. As the increasing of
hyperspectral remote sensing sensor precision and the
commercialization of hyperspectral sensors, hyperspectral
remote sensing technology plays a greater role in agriculture,
environment, military, transportation, survey, planning and any
other fields. With the advantages of large number of spectral
bands, high spectral resolution, narrow band width and high
credibility to distinguish targets, hyperspectral images is with
high dimension, large data amount, and high correlation
between bands caused by the redundancy of information. Vast
amounts of hyperspectral data brought great pressure to transfer
and storage.
At present there are three main compression methods for
hyperspectral image: the method based on transform, the method
based on vector quantization and the method based on
prediction. Several compression algorithms have been designed
for multispectral and hyperspectral images. Most of the
algorithms are based on decorrelating transforms, in order to
exploit spatial and spectral correlation. Several methods that
treat differently spectral and spatial redundancy have been
investigated. Principal component analysis method is widely
used in hyperspectral image compression to spectral
correlation[2]-[3]. In [16] vector quantization (VQ) and
Karhunen-Loeve transform (KLT) on the spectral dimension is
used to exploit the correlation between bands and discrete
wavelet transform is used in the spatial domain. In [14] spectral
decorrelation is performed using the Karhunen-Loeve Transform
and the 9-7 wavelet transform as part of the JPEG-2000 process.
In [2] the Karhunen–Loeve transform (KLT) is used to
decorrelate the data in the spectral domain, followed by a two-
dimensional(2-D) discrete cosine transform (DCT) in the spatial
domain. In [3] low-complexity KLT, along with a hybrid
wavelet based scheme, have been integrated into a JPEG 2000
Part 2 compliant scheme. In [15] a hybrid three-dimensional
wavelet transform for spectral and spatial decorrelation in the
framework of Part 2 of the JPEG2000 standard is employed. 3D-
SPIHT and 3D-SPECK algorithm for hyperspectral images have
been used for the same purpose in [10]-[11]. Context-based
adaptively lossless coding method is proposed in the literature
[9]. Without considering the correlation between spectrum the
compression performance is not good enough.
In this paper, we investigate the performance of PCA for
spectral decorrelation in conjunction with CCSDS based on
DLWT. First, we use principal component analysis method to
extract the background information from hyperspectral image[2].
Second, by using CCSDS based on DLWT, We compress the
residual image whose background message has been separated.
Our experiments indicate that better performance occurs when
our directional lifting scheme is applied on CCSDS.
2. DLWT BASED SPATIAL DECORRE-
LATION FOR HYPERSPECTRAL IMA-
GES
2.1 Directional Lifting Wavelet Transform
Two-dimensional DLWT [12]-[13] is used in the proposed
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that
copies bear this notice and the full citation on the first page. Copyrights
for components of this work owned by others than ACM must be
honored. Abstracting with credit is permitted. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee. Request permissions
from Permissions@acm.org.
ICIMCS '15, August 19-21, 2015, Zhangjiajie, Hunan, China
© 2015 ACM. ISBN 978-1-4503-3528-7/15/08…$15.00
DOI: http://dx.doi.org/10.1145/2808492.2808525