Elsevier

Neurocomputing

Volume 148, 19 January 2015, Pages 561-568
Neurocomputing

SAR complex image data compression based on quadtree and zerotree Coding in Discrete Wavelet Transform Domain: A Comparative Study

https://doi.org/10.1016/j.neucom.2014.07.007Get rights and content

Highlights

  • SAR complex image data compression based on wavelet-quadtree is proposed.

  • QC-DWT has achieved the-state-of-the-art performance.

  • QC-DWT has achieved higher performance compared with wavelet-zerotree.

Abstract

A SAR complex image data compression algorithm based on quadtree coding (QC) in discrete wavelet transform (DWT) domain (QC-DWT) is proposed. We show that QC-DWT achieves the best performance for SAR complex image compression. Besides this, in this work, we observed a novel phenomenon that QC-DWT outperforms the zerotree based wavelet coding algorithms, e.g., Consultative Committee for Space Data Systems-Image Data Compression (CCSDS-IDC) and Set Partitioning in Hierarchical Trees algorithm (SPIHT) for SAR complex image data, and there exists deficiency of CCSDS-IDC for SAR complex image data compression. This is because the DWT coefficients of SAR complex image data always have intrascale clustering characteristic and no interscale attenuation characteristic, which is different from that of SAR amplitude images and other optical images.

Introduction

Synthetic aperture radar (SAR) has been widely used in remote sensing for both civil and military applications. Unlike optical image, SAR image data is always complex. For example, interference SAR can use the phase difference of two SAR complex image data to obtain the elevation information and has been widely applied in the environmental monitoring, mapping, and other fields [1]. However, vast amounts of SAR complex image data require transmission and storage resources, which raise the needs for efficient SAR complex image data compression.

DWT based image coding is the representative coding algorithm for SAR image data compression [2], [3], [4]. Image wavelet coefficients often exhibit attenuation and clustering characteristics [5]. Accordingly, two types of image coding algorithms are popular. The first mainly uses the attenuation characteristic, such as Set Partitioning in Hierarchical Trees algorithm (SPIHT) [6] and Consultative Committee for Space Data Systems-Image Data Compression (CCSDS-IDC) [7], which have already been used for SAR amplitude image data compression [4], [7]. The second mainly uses the clustering characteristic, such as quadtree coding (QC) [8], which has also been used for SAR amplitude image compression [3]. The wavelet image coding algorithms exploiting these two characteristics show similar performance for SAR amplitude and optical images [7], [9].

For SAR complex image data compression, CCSDS-IDC based on DWT (DWT-CCSDS), which exploits the wavelet attenuation characteristic, exhibits poor performance for SAR complex image data due to the lack of interscale attenuation property [2]. Based on this, DLWT-CCSDS which combines directional lifting wavelet transform (DLWT) with the zerotree based bit plane encode (BPE) algorithm of CCSDS-IDC is proposed. It outperforms the DWT-CCSDS because DLWT can concentrate the energy to low frequency, which is beneficial to zerotree coding. However, for the test images, the K-term nonlinear approximation of DLWT is not as good as that of DWT [2], which limits further performance improvement based on DLWT for SAR complex image data. On the other hand, although QC, which exploits the intrascale clustering property, has already been used for SAR amplitude image compression [3], it has not been used for SAR complex image data compression.

In this work, to achieve a SAR complex image data compression algorithm with high performance, we first analyze the DWT coefficients’ intrascale and interscale characteristics of SAR complex image data in Section 2. Then a QC based SAR complex image data compression algorithm is proposed in Section 3. Experimental results are shown in Section 4. Section 5 concludes this paper.

Section snippets

DWT coefficients characteristics analysis of SAR complex image data

The SAR complex image data used in this work consists of 1024×1024 pixel with 16-bit per pixel, which is downloaded from the U.S. Sandia National Laboratories [10]. Fig. 1 shows the real-parts of the SAR complex images, and the imaginary-parts and amplitude images are similar to the real-parts and thus are omitted. Here, we adopt three-level 9/7 biorthogonal DWT and analyze the properties of SAR complex image data wavelet coefficients from two aspects: intrascale and interscale correlations.

Quadtree coding for SAR complex image data

Since the wavelet coefficients exhibit clustering property, we adopt QC to encode SAR complex image data. The basic idea is to continuously divide the image into four parts, and check the significance of each part. Before giving the QC, we define some variables: (1) T: quantization level; (2) LSP: the list of significant coefficients which is used to store the coordinate of significant coefficients; (3) LIB: the list used to store the up and bottom coordinates of the significant blocks; (4) sτ(k

Coding performance of quadtree and zerotree coding for SAR complex image data

We use amplitude peak signal-to-noise ratio (PSNR) and mean phase error (MPE) [2] to measure the performance of coding algorithms for SAR complex image data. Here, the real- and imaginary-parts of SAR complex image data are encoded with equal rate separately. Moreover, we adopt three-level 9/7 biorthogonal DWT decomposition in all experiments. QC based on DWT (DWT-QC) and zerotree coding algorithm: DWT-CCSDS are used for comparison. Since SPIHT algorithm shows similar performance to that of

Conclusions

In this work, we have proposed DWT-QC for SAR complex image data compression. Experimental results have shown that, for SAR complex image data, DWT-QC can achieve higher performance than zerotree coding algorithms. For SAR amplitude image data, QC achieves similar performance compared with zerotree coding algorithms. We also find CCSDS-IDC suffers low efficiency for SAR complex image data compression. This is because, for SAR complex image data, clustering characteristic exists in wavelet

Acknowledgements

This work was supported by National Natural Science Foundation of China (no. 61373113) and the Fundamental Research for the Central University (no. xjj2012023).

Xingsong Hou received the Ph.D. degree from Xi’an Jiaotong University, Xi’an, China, in 2005. Now, he is an Associate Professor with the School of Electronics and Information Engineering, Xi’an Jiaotong University. His research interests include video/image coding, wavelet analysis, sparse representation, sparse representation and compressive sensing, and radar signal processing. During October 2010–2011, he was a Visiting Scholar at Columbia University, New York, USA.

References (11)

  • P.A. Rosen, S. Hensley, I.R. Jourhin, F.K. Li, S.N.Madsen, E. Rodríguez, R.M. Goldstein, Synthetic aperture radar...
  • X.S. Hou et al.

    Complex SAR image compression based on directional lifting wavelet transform with high clustering capability

    IEEE Trans. Geosci. Remote Sens.

    (2013)
  • X.S. Hou et al.

    SAR image data compression using wavelet packet transform and universal-trellis coded quantization

    IEEE Trans. Geosci. Remote Sens.

    (2004)
  • Z. Zeng et al.

    SAR image data compression using a tree-structured wavelet transform

    IEEE Trans. Geosci. Remote Sens.

    (2001)
  • M.S. Crouse et al.

    Wavelet-based statistical signal processing using hidden Markov models

    IEEE Trans. Signal Proces.

    (1998)
There are more references available in the full text version of this article.

Cited by (16)

  • Robot Systems for Rail Transit Applications

    2020, Robot Systems for Rail Transit Applications
  • Quadtree coding with adaptive scanning order for space-borne image compression

    2017, Signal Processing: Image Communication
    Citation Excerpt :

    The tree-based partitioning methods are also applied for space-borne equipments. In [23], a SAR complex image data compression algorithm based on QC in wavelet transform domain is proposed, showing that QC achieves the best performance for SAR complex image compression. In [24], the quadtree in QC is replaced by the binary tree, which achieves better performance.

  • Remote sensing image compression based on binary tree and optimized truncation

    2017, Digital Signal Processing: A Review Journal
    Citation Excerpt :

    In [18], a three dimensional SPECK was proposed for hyperspectral image compression. In [19], a SAR complex image data compression algorithm based on QTC in wavelet transform domain was proposed, showing that QTC achieves the best performance for SAR complex image compression. In [20], the cubic B—spline fuzzy transform was proposed for an efficient and secure compression in wireless sensor networks.

  • Lossless image compression based on integer Discrete Tchebichef Transform

    2016, Neurocomputing
    Citation Excerpt :

    Lossy coding which provides great compression gains at the expense of information integrity has been widely used in digital camera, World Wide Web, mobile device and so on. On the contrary, lossless coding that holds the information integrity throughout the entire encoding and decoding process has been introduced as a necessary procedure in the application of medical images [1], remote sensing [2,3], satellite communications [4], etc. Transform coding techniques such as Karhunen-Loeve transform (KLT) [5], discrete cosine transform (DCT) [6] and Discrete Wavelet transform (DWT) [7,8] are very important tools in image and video compression.

View all citing articles on Scopus

Xingsong Hou received the Ph.D. degree from Xi’an Jiaotong University, Xi’an, China, in 2005. Now, he is an Associate Professor with the School of Electronics and Information Engineering, Xi’an Jiaotong University. His research interests include video/image coding, wavelet analysis, sparse representation, sparse representation and compressive sensing, and radar signal processing. During October 2010–2011, he was a Visiting Scholar at Columbia University, New York, USA.

Min Han received the B.S. degree from Xi’an University of Post & Telecommunications in 2011. Now she is pursuing her M.S. degree from Xi’an Jiaotong University, Xi’an, China. Her research interest is image coding.

Chen Gong received the B.S. degree in electrical engineering and mathematics (minor) from Shanghai Jiaotong University, Shanghai, China in 2005, and M.S. degree in electrical engineering from Tsinghua University, Beijing, China in 2008. He received the Ph.D degree in Electrical Engineering from Columbia University in May 2012. Now he is working in Qualcomm Inc., San Diego, Californina, USA. His research interests are in the area of channel coding and modulation techniques for wireless communications and signal processing.

Xueming Qian received the B.S. and M.S. degrees from the Xi’an University of Technology, Xi’an, China, in 1999 and 2004, respectively, and the Ph.D. degree from Xi’an Jiaotong University, Xi’an, China, in 2008. He was awarded a Microsoft fellowship in 2006. From 1999 to 2001. He was an Assistant Engineer at Shannxi Daily. From 2008 until now, he has been a Faculty Member with the School of Electronics and Information Engineering, Xi’an Jiaotong University. He was a Visiting Scholar at Microsoft Research Asia from August 2010 to March 2011. His research interests include video/image analysis, indexing, and retrieval.

View full text