Blind Detection of LSB Watermarking at Low Embedding Rate in Grayscale Images

Without any knowledge of original image a novel blind detection method on existence of LSB (Least Significant Bit) watermarking in grayscale images at low embedding rate is proposed in this paper. The proposed method employs image segmentation as preprocessing. Having studied the local character of MST (minimum spanning tree) of components the entropy of image is defined as statistical variable. The test images are grouped into two groups according to whether they are compressed or not. Compared with RS method and SPA method the proposed method gives better detection results at low embedding rate and different probability of false positive.