Image Segmentation in Wavelet Domain using Fuzzy L
no vote
Three novel algorithms: WK-Means, W-FCM and WCPSFCM, which employ segmentation procedure in wavelet transform space, are developed in this work. All of these procedures are compared with analogues methods without WT application. The commonly used criteria (AUC and ROC curve based on sensibility and specificity characteristics have been used in performance analyses, confirming the better properties of the proposed algorithms. According to experimental verification, the novel WCPSFCM framework, which employs an additional procedure to find a number of clusters in automatic mode without the intervention of a specialist, performs sufficiently well in segmentation.