Abstract: In order to extract the cellulose of Miscanthus sacchariflorus cells, the fast discrete curvelet transform combi- ning with gray level co-occurrence matrix was applied to extract the image feature vectors, and then SVM was used to segment the images. Experiments arc carried out with this method, and the results show that it is effective to compute the gray level co-occeurrence matrix of sub-band coefficient with”Angle second order moment"，”contrast"，”correla- tion" and”entropy" as the four statistics in image transform domain. According to this, texture segmentation on Miscanthus sacchariflorus cells is feasible. And compared with the texture segmentation method based on gray level co- occurrence matrix(ULCM) , the new method shortens the running time and makes the segmentation accuracy improved.

Key words: Miscanthus sacchariflorus, Ccllulosc, Curvclct, GLCM, SVM