کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
528886 | 869616 | 2013 | 10 صفحه PDF | دانلود رایگان |

• A sector-wise (512 bytes) JPEG fragment classification approach is proposed.
• Using DCT coefficient analysis to directly extract visual inconsistency features.
• The approach is conducted before the iDCT and simultaneously with JPEG decoding.
• New DCT coefficient analysis methods are proposed to measure directional emphasis.
In this paper, we propose a sector-wise JPEG fragment classification approach to classify normal and erroneous JPEG data fragments with the minimum size of 512 bytes per fragment. Our method is based on processing each read-in sector of 512 bytes with using the DCT coefficient analysis methods for extracting the features of visual inconsistencies. The classification is conducted before the inverse DCT and can be performed simultaneously with JPEG decoding. The contributions of this work are two-folds: (1) a sector-wise JPEG erroneous fragment classification approach is proposed (2) new DCT coefficient analysis methods are introduced for image content analysis. Testing results on a variety of erroneous fragmented and normal JPEG files prove the strength of this operator for the purpose of forensics analysis, data recovery and abnormal fragment inconsistencies classification and detection. Furthermore, the results also show that the proposed DCT coefficient analysis methods are efficient and practical in terms of classification accuracy. In our experiment, the proposed approach yields a false positive rate of 0.32% and a true positive rate of 96.1% in terms of erroneous JPEG fragment classification.
Journal: Journal of Visual Communication and Image Representation - Volume 24, Issue 7, October 2013, Pages 857–866