ADQ2.md

Module pyIFD.ADQ2

This module provides the ADQ2 Algorithm

Functions

ceil2(x1) : Applies ceil to vector x1, but if an element is close to an integer, it is raised by 0.5.

Args:
    x1: Input vector

Returns:
    x2: Output ceiling vector

floor2(x1) : Applies floor to vector x1, but if an element is close to an integer, it is lowered by 0.5.

Args:
    x1: Input vector

Returns:
    x2: Output floor vector

getJmap(impath, ncomp=1, c1=1, c2=15) : Main driver for ADQ2 algorithm.

Args:
    impath: Input image path, required to be JPEG with extension .jpg
    ncomp: index of color component (1 = Y, 2 = Cb, 3 = Cr)
    c1: first DCT coefficient to consider (1 <= c1 <= 64)
    c2: last DCT coefficient to consider (1 <= c2 <= 64)

Returns:
    maskTampered: estimated probability of being tampered for each 8x8 image block. Equivalent of OutputMap
    q1table: estimated quantization table of primary compression
    alphatable: mixture parameter for each DCT frequency

Todos:
    * Check returns necessary

ibdct(a, n=8) : Performs an inverse discrete cosine transorm on array a with blocks of size nxn.

Args:
    a: Array to be transformed. (2d array)
    n (optional, default=8): Size of blocks.

Returns:
    b: Output after transform. (2d array)

jpeg_rec(image) : Simulate decompressed JPEG image from JPEG object.

Args:
    image: JPEG object. (jpegio struct).

Returns:
    IRecon: Reconstructed BGR image
    YCbCr: YCbCr image