首页 | 本学科首页   官方微博 | 高级检索  
     检索      

两种新的光照方向估计方法
引用本文:杨杰,邓志鹏,郭英凯,李介谷.两种新的光照方向估计方法[J].上海交通大学学报,2002,36(6):894-896.
作者姓名:杨杰  邓志鹏  郭英凯  李介谷
作者单位:上海交通大学,图象处理与模式识别研究所,上海,200030
基金项目:国家自然科学基金资助项目 (3 0 170 2 74)
摘    要:光照方向估计是计算机视视的重要研究内容,提出了两种新的光照方向估计方法:改进的Pentland方法和加权平均法,前者修正了Pentland方法在数学上的缺陷,利用了图像中不同区域对光照方向估计的贡献,其性能比原方法有较大改进,后者利用图像中奇点的作用,并且考虑了图像中不同地形区域的特点,是一种稳定精确的光照估计方法,通过实验证明了它们的有效性。

关 键 词:计算机视觉  光照方向估计  图像处理  Pentland方法  加权平衡法  图像微分
文章编号:1006-2467(2002)06-0894-03

Two New Approaches for Illuminant Direction Estimation
YANG Jie,DENG Zhi peng,GUO Ying kai,LI Jie gu.Two New Approaches for Illuminant Direction Estimation[J].Journal of Shanghai Jiaotong University,2002,36(6):894-896.
Authors:YANG Jie  DENG Zhi peng  GUO Ying kai  LI Jie gu
Abstract:Illuminant direction estimation is an important research content of computer vision. The paper analysed the current approaches for illuminant direction estimation, presented two improved approaches for illuminant direction estimation: improved Pentland's approach and approach of weighted mean. The former has overcome some drawbacks of Pentland's approach in mathematics. It makes full use of the different areas in an image which contribute illuminant direction estimation and improves the performance of the original approach. The latter makes full use of the effect of singular points in an image, and considers the characteristics of different landforms in the image, so it is a stable and accurate approach for illuminant direction estimation. The effect of those two new approaches was proved by experiments.
Keywords:computer vision  illuminant direction estimation  image processing
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号