Abstract:Detection of double JPEG compression plays an important role in digital image forensics. Some successful approaches have been proposed to detect double JPEG compression when the primary and secondary compressions have different quantization matrices. However, detecting double JPEG compression with the same quantization matrix is still a challenging problem. In this paper, an effective error-based statistical feature extraction scheme is presented to solve this problem. First, a given JPEG file is decompressed to form a reconstructed image. An error image is obtained by computing the differences between the inverse discrete cosine transform coefficients and pixel values in the reconstructed image. Two classes of blocks in the error image, namely, rounding error block and truncation error block, are analyzed. Then, a set of features is proposed to characterize the statistical differences of the error blocks between single and double JPEG compressions. Finally, the support vector machine classifier is employed to identify whether a given JPEG image is doubly compressed or not. Experimental results on three image databases with various quality factors have demonstrated that the proposed method can significantly outperform the state-of-the-art method.

Chinese Textual Image Compression Based on Multi-feature Extraction

Double-Compression Detection in JPEG Steganalysis

Double-compressed JPEG images steganalysis with transferring feature

Double Compression Detection Based on Markov Model of the First Digits of DCT Coefficients

Image Classification Method by Combining Multi-features and Sparse Coding

Support Vector Machine With Adaptive Parameters In Image Coding

Jbig2 Text Image Compression Based on Ocr

A Compression Algorithm for English-Text Based on Dynamic Dictionary

Texture Clustering Combining Complex Wavelet Features and Local Binary Pattern

Lossy JBIG2 Based on Optical Character Recognition

An Efficient Compression Coding System for Motion Images

A feature compression method based on similarity matching

A Study of Image Compression Employing Multi-wavelets

Deep Network-Based Image Coding for Simultaneous Compression and Retrieval

Multi-Modality Deep Network for Extreme Learned Image Compression

An Improved Image Compression Method Using Cartoon-Texture Decomposition

A Lossless Compression Method for Multi-Component Medical Images Based on Big Data Mining

An Effective Method for Detecting Double JPEG Compression with the Same Quantization Matrix.

A Cnn-Based Image Compression Scheme Compatible With Jpeg-2000

Codedretrieval: Joint Image Compression and Retrieval with Neural Networks.

Chinese Text Detection and Location for Images in Multimedia Messaging Service