摘要
织物疵点图像的消噪是疵点识别和分类的重要预处理步骤。采用中值滤波、Wiener滤波和小波阈值化消噪3种方法对织物疵点图像进行消噪处理。在采用中值滤波和Wiener滤波时,同时选用3×3和5×5滤波器进行消噪;在采用小波阈值化消噪时,计算图像全局阈值,同时采用软、硬阈值消噪方法,对疵点图像进行消噪。通过实验比较,采用小波阈值化方法消噪时,疵点图像边缘清晰,峰值信噪比显著提高,其效果明显好于中值滤波和Wiener滤波;采用小波阈值化消噪后的疵点图像可在特征提取和识别中使用。
The image denoising is the important process of fabric defects identification and classification. This paper includes three image denoising methods, which are median value filtering method, Wiener filtering method and denoising with the wavelet thresholding. Both of 3 × 3 and 5 × 5 filters are used when median value filtering and Wiener filtering methods are implemented. The global threshold value is computed and the soft and hard thresholding methods are both executed when utilizing wavelet thresholding. By experiment and comparison, we find that the edge of fabric defects is clearer and the PSNR is improved distinctly when denoised by wavelet thresholding method, the performance of which is much better than that of median value filtering method or Wiener filtering method; the fabric defects image denoised with wavelet thresholding method can be used for features extracting and defects identification.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2007年第11期128-131,135,共5页
Journal of Textile Research
关键词
织物疵点
图像消噪
中值滤波
WIENER滤波
小波变换
fabric defect
image denoising
median value filtering
Wiener filtering
wavelet transform
