Abstract: A new multi-feature-based blind steganalysis scheme for JPEG images using sharpening filter is presented. In this scheme, the Markov matrix is extended to distill the intra-block correlations and the inter-block correlations; the sharpening filter is adopted to enhance the image edge; the Jessica's calibration technique is used to estimate original image and construct feature vectors; and the support vector machine classifier is applied to get results at last. Many images are tested and compared. Experimental results show that our method do better than others at the low embedding rate.