Image segmentationmethods based on spectral clustering overcome some drawbacks of the so-called central-grouping. Nyström is one of them,which uses only a partial smaller set of samples to replace the whole image pixels. In order to utilize the region information and select the sample set of the image, an image segmentation algorithm based on superpixel and improved Nyström algorithm is proposed in this paper. Firstly, region information is obtained by the superpixel method. Then the similarity measure for regions is constructed. Finally, an interval sample strategy is designed in Nyström and the regions are clustered to create the image segmentation result. With this method, the instability of random sampling is overcome, and the time complexity of color image segmentation is reduced. This method is applied to some images selected from Berkeley and VOC segmentation images. Experimental results show that our method has more advantages than FCM and Nyström algorithm in segmenting images.
CITATION STYLE
Zhao, J., Liu, H. Q., & Zhao, F. (2017). Color image segmentation based on superpixel and improved Nyström algorithm. In Advances in Intelligent Systems and Computing (Vol. 510, pp. 607–615). Springer Verlag. https://doi.org/10.1007/978-3-319-46206-6_56
Mendeley helps you to discover research relevant for your work.