The Research On Double JPEG Compression Detection Algorithm Of Low Quality Compression Factor

Since the JPEG format can achieve a high compression ratio while having high quality,and has the advantage of suitable for human visual characteristics,and is the format adopted by the most image acquisition tools.The rapid development of image processing software makes it easy for anyone to modify the image.The tampering image can arrive the degree of mix the false with the genuine.Such images have caused a series of information security problems in insurance,law,news,medical and other industries.In general,the image processing software will save the forged image in JPEG format to reduce the traces left by the tampering operation.If the original image is a JPEG image,the process of saving is actually that the image undergoes JPEG compression again.That is to say,the tamper image is usually that undergoes double JPEG compression.Looking at it another way,the JPEG double JPEG compressed image is most likely a forged image,and the problem of detecting the tamper image can be converted into the problem of detecting the double compressed image.In this paper,the low-quality double JPEG compression image detection algorithm with the same quality factor is studied.The specific work content can be summarized as:1)A low-quality JPEG double compressed image detection algorithm based on low-frequency components and error images(LFEI)is proposed.The low frequency component and the error image are analyzed at first,and then extract relevant statistical features based on the analysis.Since most of the high-frequency components will be lost during JPEG compression,the lowfrequency components are more reliable.With this in mind,our method first analyzes the low-frequency DCT coefficients and lost information in single and double JPEG compressions.The statistical features are then extracted to represent the difference between the sequential two compressed versions.Finally,the statistical features are classified via the support vector machine.Experimental results based on two standard image databases verify that the proposed method is superior to the selected comparison algorithms.In this method,the average detection rate of the low-quality factors(ranging from 70 to 20)is 87.75%.2)A low-quality JPEG double compression detection algorithm based on intrinsic fingerprints(BIF)is proposed.It is well known that after the original image has been tampered,there are some inherent artifacts on the compressed image.These hard-to-removed artifacts could be used as fingerprints for the forgery detection.In the previous research work,the fingerprint for double JPEG compression is defined in terms of geometrical change,the double quantization effect,and the block artifacts.However,when the quality factor is low,such fingerprints are too sight to be effectively detected.To solve this problem,we propose to use the fingerprints of the compressed image content and the error image to complete the detection.First,three histograms are constructed for the JPEG single and double double images to represent the overall distribution of the data in the image texture information,the time domain and the frequency domain,respectively.The total variation and divergence of the histogram are used as fingerprints to describe changes in the histogram.In addition,the fingerprints on the error image makes detection more reliable.The experimental results verify that the proposed method can obtain more effective results than the traditional detection methods.