LSB steganography is a popular method for secret information embedding wherein the lowest bit plane of cover image is replaced by that of the embedded information. There are many efficient algorithms for LSB steganography detection, but the low bit rate change of cover image cannot be accurately estimated. In this paper, we proposes a novel method to tackle the problem by using the posterior probability. Firstly, the relationship between the image pixels and the pixels around representation is established, based on which the EM algorithm is adopted to estimate the posterior probability for each pixel, and then the posterior probabilities are taken as weights to the pixels, so that the WS model is applied to estimate the low bit rate change. Experimental results show that this method can effectively and accurately estimate the low bit rate change, and thus the performance of the LSB steganography detection algorithm is improved.