JPEG image passive evidence obtaining method based on sampling maximum likelihood estimation

The invention provides a JPEG image passive evidence obtaining method based on sampling maximum likelihood estimation. The method comprises steps of filtering a grayscale image by using an enhanced cross filter and decomposing an obtained an image edge detection noise spectrum in the horizontal and vertical direction; performs derivation on a horizontal component or vertical component, extracting a periodic signal from a derivation result, computes the maximum likelihood estimation B of the block size of an original image, computing the block size estimation B of the original image by using RSMLE; dividing the noise spectrum into multiple non-overlapped child blocks, estimating the block size of each child block by using the RSMLE under a constraint condition to obtain B'; determining whether the B' is the same as the B, if yes, determining that the B' is a non-tampered area, if not, determining the corresponding block of the B' is a tampered area; refining the preliminary tampered area in order to remo