A Novel Algorithm for Quantization Matrix Estimation for JPEG Error Analysis to Digital Image Forensics

Abstract

JPEG is one of the widely used image formats in digital image processing. The analysis and the discovery of JPEG Image compression methods contribute a major significant part in digital images forensic examination. In recent years, the JPEG images can be compressed several times such as single, double and multiple times, but if the compression is applied to same image again and again without estimation of the quantization results ,it becomes very complicated to detect JPEG image compression results. Since same quantization matrix is used for several number of compression process, the results of quantization has to be analyzed in each step of compression to solve the above problem.
Moreover, error values in the quantized matrix and resizing of the image samples are not carried out effectively. In order to overcome these problems, in this research work, compressed images are resized using improved seam covering method and estimation of quantization matrix results for single and double DCT compression methods using Fuzzy Neural Network (FNN) to investigate the error values.
From FNN, the quantization matrix result is examined in DCT compression methods in order to differentiate the singly and doubly compressed images, which results quantization error values in the compression methods. Wide-ranging of experiments shows that the proposed DCT compression with FNN have higher detection rate for double and single JPEG compression successfully when compared with the existing DCT compression methods extensively

Copyright © 2014 Praise Worthy Prize - All rights reserved.