|
|
|
| Mean Shift Based Adaptive Filtering and Its Applications to Spectra Signal Processing |
| Liu Rong①; Duan Fu-qing②; Liu San-yang①;Wu Fu-chao② |
①Department of Mathematics, Xidian University, Xi’an 710071, China;
②National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing 100080, China |
|
|
|
|
Abstract An adaptive bilateral filtering method based on mean shift algorithm is presented. The filter is governed by the kernel width in spatial domain, which controls the spatial extent of nearby data for filtering. Its kernel width in range domain is chosen adaptively by the local characteristic of the signal. It can remove impulsive noise and improve smoothing of non-impulsive noise with edges preserved. Comparisons with Gaussian filter and median filter were made. Applications to spectra signal processing show this method can suppress noises in spectra effectively and reduce the amount of smoothing near spectral lines.
|
|
Received: 27 July 2004
|
|
|
|
|
|
|
|
|
2006, Vol. 28 
