Unsupervised colour-texture image segmentation via nonsubsampled contourlet transform and neutrosophic set

This paper presents an efficient scheme for unsupervised colour-texture image segmentation using nonsubsampled contourlet transform and neutrosophic set. First, colour and texture information of the image are extracted via CIE Luv colour space model and nonsubsampled contourlet transform, respectively. Then, extracted colour and texture information are transformed to neutrosophic set domain efficiently by our proposed approach. In neutrosophic set based image segmentation, the indeterminacy assessment of the image in neutrosophic set domain is notified by entropy concept. So, an appropriate indeterminacy reduction operation is proposed and employed to achieve to better segmentation results. Finally, K-Means clustering is applied on the proposed method to perform segmentation of the image in which the cluster number. Results show that our segmentation Approch outperforms other methods in segmentation of images in Berkeley dataset