Combining pixon concept with wavelet thresholding in medical image segmentation

This paper presents an innovative pixon-based method for image segmentation. The novel method uses the combination of wavelet thresholding and the pixon concept. In our method the wavelet thresholding technique is used to smooth the image and prepare it for a more efficient pixon forming. In addition, utilizing the wavelet transform results in decreasing the pixons number, a faster performance and more robustness against unwanted environmental noises. In the next stage, the appropriate pixons are extracted and eventually we segment the image with the use of a hierarchical clustering method. The results of applying the proposed method on several different images indicate its better performance in image segmentation compared to the other methods.