作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (01): 213-214,217. doi: 10.3969/j.issn.1000-3428.2011.01.073

• 图形图像处理 • 上一篇    下一篇

基于改进PCNN的彩色图像混合噪声滤除

王 成,黎绍发,何 凯,涂泳秋   

  1. (华南理工大学计算机科学与工程学院,广州 510006)
  • 出版日期:2011-01-05 发布日期:2010-12-31
  • 作者简介:王 成(1977-),男,博士研究生,主研方向:图像处理,模式识别;黎绍发,教授、博士生导师;何 凯,硕士研究生;涂泳秋,博士研究生
  • 基金资助:
    广东省工业重点攻关基金资助项目(2004B10101032)

Mixed Noise Removal for Color Images Based on Improved PCNN

WANG Cheng, LI Shao-fa, HE Kai, TU Yong-qiu   

  1. (College of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, China)
  • Online:2011-01-05 Published:2010-12-31

摘要: 基于L&A-PCNN模型的彩色图像混合噪声滤除算法存在算法调试须人工干预、对彩色图像滤波易出现污迹斑等问题。针对上述不足,提出一种彩色图像混合噪声自适应滤除算法。通过理论和实验分析获得L&A-PCNN模型关键参数的自适应定义和滤波算法中图像噪点的判别方法。实验结果表明,相比L&A-PCNN算法,该算法的PSNR有9%~18%的提高,处理后图像的视觉效果更好,并具有较好的自适应性和健壮性。

关键词: 脉冲耦合神经网络, 线性衰减阈值, 混合噪声, 彩色图像 ?

Abstract: Aiming at the deficiency of mixed noise removal algorithm for color images based on L&A-PCNN, this paper proposes an adaptive algorithm of mixed noise removal for color images based on improved L&A-PCNN. The adaptive setting method for key parameters and the estimation method of image pixels blurred by noises of L&A-PCNN model are gained with theoretical and experiments analysis. Experimental results show that the algorithm improves denoising performance of PSNR from 9% to 18% than L&A-PCNN algorithm, and gets a better visual effect when dealing with color images mixed noise removal, showing great adaptability and robustness.

Key words: Pulse Coupled Neural Network(PCNN), linear-attenuated threshold, mixed noise, color image

中图分类号: