Abstract��Image retouching is a technique widespreadly used in image tampering.To implement image retouching blind detection, a blind identification algorithm for image was proposed. The algorithm firstly searches for each block in image, and inserts each block into the KD tree. Thus obtain the same value or the nearest match block, then it uses the hierarchical clustering of the position vector to eliminate the scattered matching. Finally, it applies the 7-tap Laplacian filter to count weight of zero connectivity of suspicious blocks to eliminate false positives, thereby locate accurately the tampered area. Experiments show that this method can efficiently and accurately identify the use of a class of image manipulation techniques for uncompressed images and high quality compressed images. When it is applied to the images that have higher compression levels, accurate results are also obtained if the region that has been retouched is sufficiently large.