First JPEG quantization matrix estimation based on histogram analysis

Abstract

To assess if a digital image has been (or not) doubly compressed is a challenging issue especially in forensics domain where could be fundamental clarify if, in addition to the compression at the time of shooting, the picture was decompressed (in some way) and then resaved. This is not a clear indication of forgery, but it guarantees that the image, probably, is not the original one. In this paper we propose a novel technique able to recover the coefficients of the first compression in a double compressed JPEG image under some assumptions. The proposed approach exploits how successive quantizations followed by dequantizations introduce some regularities (e.g., sequence of zero and not zero values) on the histograms of coefficient distributions that could be analyzed to recover the original compression parameters. Experimental results and comparisons with state of the art methods confirm the effectiveness of the proposed approach.

Detecting Deepfakes with Deep Learning and Gabor Filters

“…These methods were previously relied on before deep learning algorithms appeared and depended on in the detection of forgery. For example, the author of Puglisi, et al, 2013, utilizes how implementing successive quantization followed by dequantizations introduces some arranges in the sequence of zeros and non-zeros in the coefficient distributions of the histogram. This method allows for recapturing the coefficients of the first compression in a double JPEC compression.…”

Estimating JPEG compression history of bitmaps based on factor histogram

“…Recently, Lin et al [9] further improved Ye et al's method by classifying the patterns of individual frequencies into four classes, and then extracting the quantization step adaptively from the energy density spectrum (EDS) of DCT coefficient histogram and the Fourier transform of EDS. In addition, there are also some works aiming to detect doubly JPEG compressed images and estimate quantization parameters of the primary JPEG compression [14][15][16][17][18][19][20][21]. These works typically assumed that the quantized DCT coefficients can be directly decoded from the input JPEG file, or the quantization table of the last (i.e., secondary) JPEG compression is known.…”