A novel rapid region-based image segmentation algorithm is introduced in the paper. It extracts colour,texture and position features of all pixels of the image first, and partitions the image into sub-blocks followed by taking the means of pixel features inside the sub-block as the eigenvector for every sub-block. Then ,the number of clusters and the initial cluster centre are obtained by using mean-shift clustering algorithm. Finally, the modified K-means algorithm is employed to cluster for completing rapid image segmentation. Experimental results show that the new method is rapid in segmenting and stable in segmented outcomes. At the same time, it avoids over-segmentation.