改善最小均方误差估计的煤尘图像去噪.pdfVIP

第 23 卷 第 9 期 电子测量与仪器学报 Vol. 23 No. 9 2009 年 9 月 JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT · 51 · 改进最小均方误差估计的煤尘图像去噪 张 伟1,2 隋青美1 (1. 山东大学控制科学与工程学院, 济南 250061; 2. 青岛科技大学自动化与电子工程学院, 青岛 266042) 摘 要: 煤尘图像在采集和传输过程中受到了各种噪声的污染。最小均方误差估计(MMSE)去噪算法对高斯噪声有较好的 去噪效果, 提出了一种改进的最小均方误差估计(IMMSE)去噪算法, 该算法改进了广义高斯分布模型的参数估计方法, 相比目 前的其他算法, 在不降低精度的情况下减少了计算量。中值滤波对脉冲噪声有较好的去噪效果, 用自适应中值滤波(AM)代替普 通的中值滤波, 更好的保留了图像的细节, 提高了去噪效果。利用 IMMSE 和 AM 自在图像去噪方面的优势, 将两者有机地结合 起来, 提出了一种称之为 IMMSE-AM 的去噪算法。用 IMMSE-AM 对真实煤尘图像进行去噪处理, 实验结果表明, 新算法提高 了煤尘图像的去噪效果, 并且计算量较小, 能够满足对煤尘浓度实时测量的要求。 关键词: 最小均方误差估计；自适应中值滤波；图像去噪；煤尘图像 中图分类号: TP 391.41 文献标识码: A 国家标准学科分类代码: 510.4050 Image denoising of coal dust based on improved minimum mean square error estimation Zhang Wei1,2 Sui Qingmei1 (1. School of Control Science and Engineering, Shandong University, Jinan 250061, China; 2. School of Automation and Electronic Engineering, Qingdao University of Science and Technology, Qingdao 266042, China) Abstract: The coal dust image is often corrupted by all kinds of noises during acquisition or transmission. Mini- mum mean square error estimation (MMSE) has good effect on Gaussian noise denoising, while adaptive median filter has good effect on pulse noise denoising. An improved MMSE (IMMSE) which improves the parameters estimation of generalized Gaussian distribution (GGD) is proposed. Combining IMMSE with adaptive median filter based on their advantages in denoising, a new denoising algorithm (IMMSE-AM) is presented to use for coal dust image denoising. Experiment results show that IMMSE-AM improved coal dust image denoising effect with less computation quantum. It can be u