In this work, we propose a new algorithm for spectral image segmentation based on the use of a kernel matrix. An efficient multiscale method is presented for accelerating spectral image segmentation. The multiscale strategy uses the lattice geometry of images to construct an image pyramid whose hierarchy provides a framework for rapidly estimating eigenvectors of normalized kernel matrices. To prevent the boundaries from deteriorating, the image size on the top level of the pyramid is generally required to be around 75×75, where the eigenvectors of normalized kernel matrices would be approximately solved by the Nyström method. Within this hierarchical structure, the coarse solution is increasingly propagated to finer levels and is refined using subspace iteration. Experimental results have shown that the proposed method can perform significantly well in spectral image segmentation as well as speed up the approximation of the eigenvectors of normalized kernel matrices.