电信科学 ›› 2016, Vol. 32 ›› Issue (6): 73-82.doi: 10.11959/j.issn.1000-0801.2016122

• 研究与开发 • 上一篇    下一篇

纯二维5/3小波变换及其在CT图像无损压缩中的应用

徐妮妮,张明明,汪剑鸣   

  1. 天津工业大学电子与信息工程学院,天津300387
  • 出版日期:2016-06-20 发布日期:2016-04-28
  • 基金资助:
    国家自然科学基金资助项目

True 2-D 5/3 wavelet transform and its application for CT image lossless compression

Nini XU,Mingming ZHANG,Jianming WANG   

  1. School of Electronics and Information Engineering,Tianjin Polytechnic University,Tianjin 300387,China
  • Online:2016-06-20 Published:2016-04-28
  • Supported by:
    The National Natural Science Foundation of China

摘要:

小波变换在数字图像处理领域有着广泛的应用,其对图像的处理常采用行列分离处理方式,这种方式不能完全吻合人眼视觉特性。针对这一情况,构造了一种与人眼视觉特性更加吻合的纯二维小波变换处理方式。首先,由一维 5/3 小波滤波器组通过 McClellan 变换构造纯二维 5/3 小波滤波器组,并用提升格式实现;然后,用该提升格式与纯二维Lazy 小波滤波器组相嵌套的形式实现图像的纯二维5/3 小波变换。为了便于工程应用,给出了其变换规程。将纯二维5/3小波变换用于CT图像的无损压缩,实验证明:对于512 dpi×512 dpi尺寸的CT图像,纯二维5/3小波变换无损压缩效果高于二维可分离5/3小波变换,每幅图像可平均节省1 989.9 byte。

关键词: 人眼视觉系统, McClellan变换, 提升格式, 纯二维5/3小波变换, 无损压缩

Abstract:

Wavelet transform is widely used in the field of digital image processing,and it usually processes image by separating the row and column,which can’t meet the human visual system(HVS)completely.To address this situation,a kind of true 2-D 5/3 wavelet transform was constructed,which was more consistent with HVS.Firstly,1-D 5/3 wavelet filter bank constructed the true 2-D Lazy wavelet filter bank by McClellan transform,and it was implemented by lifting scheme.Then,this lifting scheme was nested with the true 2-D Lazy wavelet filter bank to implement true 5/3 wavelet transform of image.To facilitate engineering applications,the transformation procedure was given.The experiments of true 5/3 wavelet transform used in CT image lossless compression,the experiment shows that,for 512×512 size CT image true 5/3 wavelet transform has better compression effect when compared with two dimensional separable 5/3 wavelet transform and can save 1 989.9 byte per piece on average.

Key words: HVS, McClellan transform, lifting scheme, lossless compression

No Suggested Reading articles found!