考虑尺度间相关性的电缆瓷套终端红外图像去噪
发布时间：2018-11-26 19:13
【摘要】：为有效抑制图像噪声,提高电气设备红外诊断的准确性,采用基于小波系数尺度间相关性和双变量收缩函数的方法对电缆瓷套终端红外图像进行去噪.将图像进行小波分解,计算小波系数尺度间的相关系数,使用模糊c-均值聚类法对相关系数聚类,即将小波系数分为有效系数和无效系数两类.对无效小波系数直接进行置零处理,对有效小波系数使用双变量收缩函数进行处理,得到真实图像小波系数的估计值.最后,对处理得到的真实图像小波系数的估计值进行重构,便得到去噪后图像.含噪图像的去噪结果表明,运用文中方法能有效地去除红外图像中的噪声,且与使用传统软阈值方法去噪得到的图像对比,文中方法去噪后的图像信噪比更高,最小均方误差更小.
[Abstract]:In order to effectively suppress image noise and improve the accuracy of infrared diagnosis of electrical equipment, a method based on wavelet coefficient scale correlation and bivariate shrinkage function is used to de-noise infrared image of cable porcelain cover terminal. The image is decomposed by wavelet, and the correlation coefficients between scales of wavelet coefficients are calculated. The correlation coefficients are clustered by fuzzy c-means clustering method, that is to say, the wavelet coefficients are divided into effective coefficients and invalid coefficients. The invalid wavelet coefficients are directly zeroed, and the effective wavelet coefficients are processed by the bivariate contraction function, and the estimated values of the real image wavelet coefficients are obtained. Finally, the real image is reconstructed from the estimated wavelet coefficients, and the denoised image is obtained. The denoising results of noisy images show that the proposed method can effectively remove the noise in infrared images, and compared with the image denoised by the traditional soft threshold method, the signal-to-noise ratio of the image after denoising is higher. The minimum mean square error is smaller.
【作者单位】： 华南理工大学电力学院;珠海供电局;广州供电局有限公司;
【基金】：国家高技术研究发展计划(863计划)项目(2015AA050201)~~
【分类号】：TM507;TP391.41
本文编号：2359378
[Abstract]:In order to effectively suppress image noise and improve the accuracy of infrared diagnosis of electrical equipment, a method based on wavelet coefficient scale correlation and bivariate shrinkage function is used to de-noise infrared image of cable porcelain cover terminal. The image is decomposed by wavelet, and the correlation coefficients between scales of wavelet coefficients are calculated. The correlation coefficients are clustered by fuzzy c-means clustering method, that is to say, the wavelet coefficients are divided into effective coefficients and invalid coefficients. The invalid wavelet coefficients are directly zeroed, and the effective wavelet coefficients are processed by the bivariate contraction function, and the estimated values of the real image wavelet coefficients are obtained. Finally, the real image is reconstructed from the estimated wavelet coefficients, and the denoised image is obtained. The denoising results of noisy images show that the proposed method can effectively remove the noise in infrared images, and compared with the image denoised by the traditional soft threshold method, the signal-to-noise ratio of the image after denoising is higher. The minimum mean square error is smaller.
【作者单位】： 华南理工大学电力学院;珠海供电局;广州供电局有限公司;
【基金】：国家高技术研究发展计划(863计划)项目(2015AA050201)~~
【分类号】：TM507;TP391.41
【相似文献】
相关期刊论文 前1条
1 高强,赵振兵,李然,俞晓雯;基于独立分量分析的近红外图像去噪方法的研究与应用[J];中国电机工程学报;2005年22期
，本文编号：2359378
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2359378.html
