›› 2017, Vol. 30 ›› Issue (9): 12-.

• 论文 • 上一篇    下一篇

非均匀图像亮度补偿的分割方法

闻敬东,张轩雄,巨志勇   

  1. (上海理工大学 光电信息与计算机工程学院,上海 200093)
  • 出版日期:2017-09-15 发布日期:2017-11-03
  • 作者简介:闻敬东(1993-),男,硕士研究生。研究方向:模式识别和图像处理。张轩雄(1965-),男,研究员,博士生导师。研究方向:微电子机械系统等。巨志勇(1975-),男,博士。研究方向:模式识别等。
  • 基金资助:

    国家自然科学基金(81101116)

Segmentation Method for Unevenly Image Brightness Compensation

WEN Jingdong,ZHANG Xuanxiong,JU Zhiyong   

  1. (School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology, Shanghai 200093,China)
  • Online:2017-09-15 Published:2017-11-03

摘要:

为消除非均匀光照对图像分割效果的影响,提出了改进的动态型高斯同态滤波的亮度补偿新方法。将非均匀图像置于HIS色系下,在频域内对其亮度分量进行同态滤波,同时引入改进后的高斯高通滤波传递函数,并保持色调和饱和度不变,得到光照均匀的图像,进而利用Otsu算法对增强后的均匀图像进行分割。实验结果表明,文中所提方法能够使得到补偿后的图像明暗适中、视觉感好,且分割的精度与速度都较为理想。

关键词: 非均匀光照, 亮度补偿, HIS色系, 改进的动态型高斯同态滤波

Abstract:

Aiming at eliminating the effect of unevenly illumination on image segmentation, a new method of illumination compensation based on dynamic Gauss type filter is proposed. The unevenly image is in the HSI color system.,then the illumination component is filtered in frequency domain, and the improved Gauss high pass filter transfer function is introduced to it, keep the hue and saturation constant, then got the uniform image. At last use the Otsu algorithm for uniform image which was enhanced before. The experimental results show that the proposed method can make the image brightness and brightness moderate, and the visual sense is good. The accuracy of segmentation is more ideal.

Key words: unevenly illumination, illumination compensation;HSI color system, improved dynamic Gauss type filter

中图分类号: 

  • TN911.73