华南理工大学学报(自然科学版) ›› 2005, Vol. 33 ›› Issue (3): 44-49.

• • 上一篇    下一篇

基于人眼视觉特性的任意形状ROI编码

梁亚玲 杨春玲 余英林 杜明辉   

  1. 华南理工大学 电子与信息学院,广东 广州 510640
  • 收稿日期:2004-05-19 出版日期:2005-03-25 发布日期:2005-03-25
  • 通信作者: 梁亚玲(1977-),女,助教,硕士,主要从事图像处理的研究 E-mail:ylliang@scut.edu.cn
  • 作者简介:梁亚玲(1977-),女,助教,硕士,主要从事图像处理的研究

Arbitrary Shaped ROI Image Coding Based on HVS

Liang Ya-ling  Yang Chun-ling  Yu Ying-lin  Du Ming-hui   

  1. College of Electronic&Information Engineering,South China Univ.of Tech.,Guangzhou 510640,Guangdong,China
  • Received:2004-05-19 Online:2005-03-25 Published:2005-03-25
  • Contact: 梁亚玲(1977-),女,助教,硕士,主要从事图像处理的研究 E-mail:ylliang@scut.edu.cn
  • About author:梁亚玲(1977-),女,助教,硕士,主要从事图像处理的研究

摘要: 研究了等级树集分割编码方法(SPIHT)在任意形状感兴趣区域(ROI)编码中的应用.根据SPIHT算法的特性、ROI小波系数提升后的特点及人眼视觉特性,提出基于人眼视觉特性的任意形状ROI编码方法.该方法综合了一般基于尺度的方法和最大平移法的优点:可调节ROI与背景(BG)的相对重要程度;支持任意形状ROI编码.另外,与最大平移法相比,本文方法具有更好的主观视觉效果,理论分析和仿真实验都表明本文方法优于MPEG-4中规定的视频静态纹理视觉对象(VO)编码.

关键词: 图像编码, 感兴趣区域, 等级分级树, 人眼视觉特性

Abstract:

The application of Set Partitioning in Hierarchical Trees(SPIHT)to the image coding of arbitrarily shaped Region of Interest(ROI)is investigated.In this investigation,according to the characteristics of SPIHT algorithm and by considering the properties of the image wavelet coefficients when the coefficients associated with ROI are scaled up,an improved algorithm based on Human Visual System(HVS)is proposed.This algorithm possesses
the merits of the generic-scaling-based method and the maxshifi method,that is,it can adjust the relative importance between ROI and the background,and can be used for the coding of arbitrarily shaped ROI.Th e results show that the proposed method is of better subjective visual quality than the maxshifi method.Moreover,theoretical ana-lyses and simulated experiments all indicate that the proposed method is better than the Visual Object(VO)coding prescribed in MPEG-4 .

Key words: image coding, region of imerest, set partitioning in hierarchical trees, human visual system