Median filtering in the wavelet domain in image segmentation

In this paper we propose a method of applying median filtering in the wavelet domain in image segmentation when the image consists of regions hard to be segmented by the spatial domain median filter. The method transforms an image into the wavelet domain and then iteratively applies the median filter in the wavelet domain and finally transforms the result into the spatial domain. The advantage of wavelet domain median filtering is that in the wavelet domain, probabilities of encountering root images are spread over sub-band images and therefore median filtering is unlikely to encounter root images at an early stage of iterations and can generate better results as the iteration increases. Better performance is obtained in segmenting images having a Gaussian noise patter, especially when regions are close to each other in their mean values. Results from computer simulation are used to demonstrate superiority of the method.