Authentication of JPEG Images Based on Genetic Algorithms

There is an increasing number of crime cases involving computers and multimedia files such as child pornographic photos, recorded voices and CCTV videos. It is not uncommon that suspects can manage to erase the files or even destroy the whole file system before seized by the law enforcement officer. Reconstructing fragmented files is still a challenge in digital forensics for both damaged and integrity files, especially without the file system information.
\n
\nFirstly, this work focuses on JPEG files, one of the most popular photograph formats, and proposes techniques for recovering partially-damaged standalone JPEG fragments by reconstructing pseudo headers. The techniques deal with missing Huffman tables and sub-sampling factors, estimate the resolution of JPEG fragments, assess the image quality of JPEG files with incorrect quantization tables, and create quantization tables that are very close to the correct quantization tables in a reasonable amount of time. Extensive experiments with real camera pictures demonstrate that the techniques can recover standalone fragments accurately and efficiently.
\n
\nSecondly, this study focuses on JPEG file carving for, in particular, heavily fragmented JPEG files when file system information damaged. Due to the fact that, it is a critical challenge to identify encrypted or compressed unknown data streams without sufficient corresponding decoding information. To begin with, we propose a technology to display and identify the JPEG file fragmentation piece, which is a partial content of a file and consist of one or more physically consecutive data blocks, from unknown data streams. A new similarity matching metric (CED) is proposed to evaluate the difference between data blocks in large scale. By evaluating the pattern of similarity distribution in fragmented file, we can identify the occurrence of fragmentation point in advance. Then, a fragmentation piece-based _le carving methodology is presented to recover heavily fragmented JPEG files. By comparing with state-of-the-art photo recovery technical, the proposed techniques can automatically and successfully recover most of fragmented files from real case.
\n
\nMoreover, with the popularity of smart phone, voice chat of instant message (IM) applications are getting popular. However, huge amount of manpower is required to listen, analyze, and identify relevant chat files of IM apps during current forensic investigation tools and techniques. This study proposes a semi-automatic integrated framework to deal with audio forensic investigation for IM apps by applying modern technologies. The main objective is to reduce the amount of manpower in the investigation. This is the first work that applies speech to text technology in voice chat of IM apps forensic. Both text and audio features are extracted to reconstruct the dialog conversation. Experiments with real case data show that the framework is promising. And it is able to translate dialog into readable text and improve the efficiency during investigation with reconstructed conversation.

Several fragile watermarking techniques cannot detect vector quantization (VQ) attacks; the block-based approach, for example. Celik et al. (IEEE Trans. Image Processing, vol. 11, no. 6, pp. 585-595, 2002) proposed a hierarchical watermarking approach to reveal the VQ attacks. In order to improve the hierarchical watermarking approach, we present a new fragile watermarking technique based on genetic algorithms (GAs) to embed watermarks into the frequency domain of a host image. Different from existing schemes, the adjustment in the host image by GAs will achieve the embedding of fragile watermarks. The technique can provide a fundamental platform for other fragile watermarking techniques. Furthermore, the embedding of watermarks into frequency domain enhances the security in fragile watermarking

Practical steganalysis inevitably involves the necessity to deal with a diverse cover source. In the JPEG domain, one key element of the diversification is the JPEG quality factor, or, more generally, the JPEG quantization table used for compression. This paper investigates experimentally the scalability of various steganalysis detectors w.r.t. JPEG quality. In particular, we report that CNN detectors as well as older feature-based detectors have the capacity to contain the complexity of multiple JPEG quality factors within a single model when the quality factors are properly grouped based on their quantization tables. Detectors trained on multiple JPEG qualities show no loss of detection accuracy when compared with dedicated detectors trained for a specific JPEG quality factor. We also demonstrate that CNNs (but not so much feature-based classifiers) trained on multiple qualities can generalize to unseen custom quantization tables compared to detectors trained for specific JPEG qualities. Their ability to generalize to very different quantization tables, however, remains a challenging task. A semi-metric comparing quantization tables is introduced and used to interpret our results.

We present a model for generating a robust JPEG quantization table using the techniques of genetic algorithms at the JPEG image compression process. After several experiments over a range of generations, the final quantization table was obtained. The detection of the best quantization table (Q-table) using genetic algorithms with the JPEG standard is a great tool to obtain the desired quality of recovered image. This method compares the SNR of the quantization tables created during the process, and choose the one with the higher SNR for a group of natural images by the natural selection, the program also give conditions to change anytime the parameters of the program to produce better results.

Because of the prevalence of internet, the information security issues have attracted much attention in recent years. Protecting privacy of exchanging information through media becomes the subject of many studies. The Quantization in JPEG is to encode the coefficients transformed by Discrete Cosine Transform (DCT) using a Quantization Table (QT) generated by a quality factor. Quantizing DCT coefficients by different QTs can be adopted to record the secret messages. Based on the variation of QTs, we propose two schemes in frequency domain to improve a data hiding method, in terms of correct extraction rate and image quality. Experimental results show that the proposed approaches can provide better correct extraction rate, higher image quality, and satisfactory embedding capacity.

Steganalysis classifier training via minimizing sensitivity for different imaging sources

Quantisation table used in joint photographic experts group JPEG standard plays a major role in the trade-off between the compression and quality of the image. Hence, finding a quantisation table is inferred as an optimisation problem. This paper proposes a classical differential evolution CDE algorithm for finding the optimal luminance quantisation table for a target bits/pixel bpp. Also, it aims to carry out an extensive performance comparison of the CDE algorithm with the classical genetic algorithm CGA. It is observed that the CDE-based quantisation table outperforms the default JPEG and CGA-based quantisation tables in terms of mean squared error MSE and peak signal to noise ratio PSNR. Images with different complexity levels have been experimented with both CDE and CGA algorithms. The validation results show that CDE guarantees a feasible solution in faster convergence rate than CGA. The results are confirmed using the statistical hypothesis test t-test.

Reversible data hiding (RDH) techniques for JPEG images have big challenges in order to improve the capacity, quality, and flexible embedding method. In this paper, we propose a new RDH algorithm using combination of selected DCT coefficients strategy with quantization tables. In order to further reduce the distortion, we survey the effect of quantized process on DCT coefficients for making the strategy of selection the embedding DCT coefficients. Compared to the conventional method, our experimental results show that our proposed method has better performance in terms of both the increasing capacity and improving the quality of embedded JPEG images after data embedding.KeywordsReversible data hiding (RDH)JPEG imageDC coefficientsAC coefficientsQuantization tables

A novel technique for data hiding in JPEG images is proposed. The technique relies on modification of zero quantised coefficients in each image block, in contrast to most previously proposed methods which also affect non-zero coefficients and/or the quantisation tables. Both embedding and extraction are performed on a per-block basis, without the need to preprocess the whole image. The proposed technique provides significantly lower distortion for similar embedding capacity, compared with recent state-of-the-art techniques.

Image compression is a basic need for the progressive transmission or storage of image data via any media. For the effective processing of image data the coding standards were developed. The most commonly used standard in current scenario is the JPEG and JPEG-2000 coding. These standards have become the industry standards and are progressively been in use for almost all image processing applications. In the process of image coding various sub-processing operations occur which process the image blocks by transformation, quantization and encoding before transmitting or storing. In the process, quantization is observed to be a time consuming and is needed to be processed in high precision so as to achieve less processing error and also in reasonable time. The process of quantization has shown greater efficiency in vector quantization than earlier methods. Vector quantization is observed to be an efficient block based lossy compression technique. Due to simpler coding process and high compression ratio VQ are widely used. Predictive vector Quantization (PVQ) is commonly used memory based vector Quantization method. The main drawback with PVQ is the high computational complexity which limits PVQ to use in real time applications. Shorter coding time can be archived by reducing the number of code words or instructions, but the reconstructed image quality will degraded in the case of fewer code words. Hence, it is necessary to propose fast coding algorithms for PVQ. In this paper a modified version of the PVQ approach has been suggested in order to process faster, which ensue to be a compatible mechanism for real time image coding in JPEG based coding system. Keywords: Compression, quantization, PVQ encoder, mean-square-error, Discrete Cosine Transform.

JPEG compression is one of image degradations that often occurs in image storing and retouching process. Estimating JPEG compression degradation property is important for JPEG deblocking algorithm and image forensic analysis. JPEG degradation exists not only in JPEG file format but also in other image formats because JPEG distortion remains after converting to another image format. Moreover, JPEG degradation property is not always uniform within an image in case that the image is collaged from different JPEG-compressed photos. In this paper, pixelwise detection of JPEG-compression degradation and estimation of JPEG quality factor using a convolutional neural network is proposed. The proposed network outputs an estimated JPEG quality factor map and a compression flag map from an input image. Experimental results show that the proposed network successfully infers the quality factors and discriminates between non-JPEG-compressed images and JPEG-compressed images. We also demonstrate that the proposed network can spot a collaged region in a fake image which is comprised of images that have different JPEG compression properties. Additionally, the network reveals that image datasets Set5 and Set14, often used to evaluate super-resolution algorithms, contain JPEG-compressed low quality images, which are inappropriate for such evaluation.

Form error evaluation of geometrical products is a nonlinear optimization problem, for which a solution has been attempted by different methods with some complexity. A genetic algorithm (GA) was developed to deal with the problem, which was proved simple to understand and realize, and its key techniques have been investigated in detail. Firstly, the fitness function of GA was discussed emphatically as a bridge between GA and the concrete problems to be solved. Secondly, the real numbers-based representation of the desired solutions in the continual space optimization problem was discussed. Thirdly, many improved evolutionary strategies of GA were described on emphasis. These evolutionary strategies were the selection operation of ‘odd number selection plus roulette wheel selection’, the crossover operation of ‘arithmetic crossover between near relatives and far relatives’ and the mutation operation of ‘adaptive Gaussian’ mutation. After evolutions from generation to generation with the evolutionary strategies, the initial population produced stochastically around the least-squared solutions of the problem would be updated and improved iteratively till the best chromosome or individual of GA appeared. Finally, some examples were given to verify the evolutionary method. Experimental results show that the GA-based method can find desired solutions that are superior to the least-squared solutions except for a few examples in which the GA-based method can obtain similar results to those by the least-squared method. Compared with other optimization techniques, the GA-based method can obtain almost equal results but with less complicated models and computation time.

A high capacity lossless data hiding scheme for JPEG images

Purpose The role of image compression is inevitable in telemedicine for the transfer of medical images. The diagnostic quality of the reconstructed image plays a vital role, and hence, the choice of compression algorithm is necessary. The medical images are analysed by the physicians for the diagnosis of anomalies. Procedures Linde Buzo Gray (LBG) is a classical algorithm for vector quantization; it produces a local optimum codebook that results in low compression efficiency. The efficiency of the vector quantization (VQ) relies on the appropriate codebook; hence, many research works based on optimization have developed for the generation of global codebook. This paper couples LBG with BAT optimization algorithm which generates an appropriate codebook. The optimization algorithm is employed not only for codebook design but also for the codebook size selection. Results The proposed BAT-LBG with dynamic codebook selection was compared with the classical LBG-VQ, JPEG lossless and BAT-LBG with static codebook selection. The algorithms were tested on abdomen CT images of five datasets, and performance evaluation was done by performance metrics like peak signal-to-noise ratio (PSNR) and mean square error (MSE). Conclusions The BAT-LBG compression with dynamic codebook size selection was found to produce efficient results when compared with JPEG lossless, classical LBG-VQ and BAT-LBG with static codebook selection. The quality of the reconstructed image was found to be good for BAT-LBG with dynamic codebook selection, and hence, it is efficient for transfer of medical images in telemedicine.

Current image re-sampling detectors can reliably detect re-sampling in JPEG images only up to a Quality Factor (QF) of 95 or higher. At lower QFs, periodic JPEG blocking artifacts interfere with periodic patterns of re-sampling. We add a controlled amount of noise to the image before the re-sampling detection step. Adding noise suppresses the JPEG artifacts while the periodic patterns due to re-sampling are partially retained. JPEG images of QF range 75-90 are considered. Gaussian/Uniform noise in the range of 28-24 dB is added to the image and the images thus formed are passed to the re-sampling detector. The detector outputs are averaged to get a final output from which re-sampling can be detected even at lower QFs. We consider two re-sampling detectors - one proposed by Poposcu and Farid [1], which works well on uncompressed and mildly compressed JPEG images and the other by Gallagher [2], which is robust on JPEG images but can detect only scaled images. For multiple re-sampling operations (rotation, scaling, etc) we show that the order of re-sampling matters. If the final operation is up-scaling, it can still be detected even at very low QFs.