Journal of South China University of Technology (Natural Science Edition) ›› 2009, Vol. 37 ›› Issue (4): 31-36.

• Computer Science & Technology • Previous Articles     Next Articles

Mixed-Noise Removal Based on L＆A-PCNN

Tu Yong-qiu1  Li Shao-fa1  Wang Cheng1  Wang Min-qin2   

  1. 1. School of Computer Science and Engineering, South China University of Technology, Guangzhou 510640, Guangdong, China; 2. School of Computer Science, Zhaoqing University, Zhaoqing 526061, Guangdong, China
  • Received:2008-05-06 Revised:2008-07-07 Online:2009-04-25 Published:2009-04-25
  • Contact: 涂泳秋（1980-），女，博士，主要从事数字图像处理、模式识别研究． E-mail:tu．yongqiu@mail．scut．edu．cn
  • About author:涂泳秋（1980-），女，博士，主要从事数字图像处理、模式识别研究．
  • Supported by:

    国家自然科学基金资助项目（60573019）

PDF

913

Abstract:

As the existing pulse-coupled neural network （PCNN） suh in blur patch and information loss during image smoothing, a models are of complex threshold functions and may remodified PCNN model L＆A-PCNN with linear-attenuated threshold and weighted average gray level output is designed. The optimal value ranges of the key parameters of the new model are then determined via mathematical reasoning and experiments. Moreover, the mixed noise which is difficult to denoise is recovered by combining the L＆A-PCNN model with a median filter. Simulated results show that the denoising performance of the new algorithm improves by 5% -30%, as compared with the existing algorithms.

Key words: pulse-coupled neural network model, linear-attenuated threshold, firing pixel, weighted average, mixed noise, median filtering