원문정보
초록
영어
The double quantization effect of JPEG provides important clue for detecting image tampering. Whenever an original JPEG image has undergone a localized tampering and saved in JPEG again, the DCT coefficients of the areas without tampering will be compressed for twice while the tampered areas only suffered once. The Alternating Current (AC) coefficient distribution accord with a Laplace probability density distribution described with parameter. This paper proposed a new double compression probability model of JPEG image to describe the change of DCT coefficients’ statistical properties after the double compression. According to Bayes’ theorem, using the posterior probability, the model can also show the eigenvalues of the double and single compressed block. We assign a dynamic adaptive threshold for the eigenvalues with the Particle Swarm Optimization Algorithm. Then the tampered region is detected and separated automatically by using the threshold. The experimental results show that the method can detect and separate the tamped area effectively and it outperforms other algorithms in terms of the detection result especially when the second compression factor is smaller than the first one. Compared with other traditional methods, the proposed approach could effectively separate the tampered regions from the tampered image without respect to the location, size and number of tampered images.
목차
1. Introduction
2. JPEG Compression Principle
3. The Mathematical Model of JPEG Tampered Image
4. The Double Quantization Effect in JPEG Compression
5. The Detection Algorithm Based on the Double Quantization Effect on DCT Coefficients
5.1. Double Quantization in Tampered JPEG Images
5.2. Extraction the Eigenvalues of the Tampered Block Based on Bayesian
5.3. Adaptive Multi-Threshold Set by Particle Swarm Optimization (PSO) Algorithm Automatically Extracts the Tampered Regions
6. Experimental Results and Analysis
7. Concluding Remarks
References