Application of an Improved Grab Cut Method in Tongue Image Segmentation

Abstract

Grab Cut is an image segmentation method based on graph theory, and it is an improved algorithm of Graph Cut. Color images can be segmented by Grab cut. However, Grab Cut has the disadvantage of long segmentation time consuming. The application of SLIC (simple linear iterative clustering) super pixel method can reduce the time consumption. According to the particularity of the larger R value in the pixel of the tongue image, the formula of SLIC color space distance is improved, so that the super pixel produced by SLIC is more suitable for tongue image segmentation. The segmentation experiment on 300 tongue images shows that the segmentation accuracy of the improved algorithm is over 0.95, and the segmentation time is reduced greatly compared with the original Grab Cut algorithm. The algorithm can reduce the time of the tongue segmentation and improve the efficiency of the tongue segmentation, while maintaining the accuracy of the segmentation.