A new approach for jpeg fragmentation point detection using sequential difference by segment (SDbS)

• Main Author: T Azmi, Tengku Norsuhaila
• Format: Thesis
• Language: English; English; English
• Published: 2019
• Subjects: TK5101-6720 Telecommunication. Including telegraphy, telephone, radio, radar, television
• Online Access: http://eprints.uthm.edu.my/676/1/24p%20TENGKU%20NORSUHAILA%20T%20AZMI.pdf; http://eprints.uthm.edu.my/676/2/TENGKU%20NORSUHAILA%20T%20AZMI%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/676/3/TENGKU%20NORSUHAILA%20T%20AZMI%20WATERMARK.pdf; http://eprints.uthm.edu.my/676/

Digital Forensic is a computer science application and an investigation means, aimed at the prevention of illegal crimes by tracking and retrieving evidence left by cyber criminals. Image recovery is one of the techniques to obtain digital evidences. This technique works by recovering damaged or corrupted digital evidences either through traditional data recovery or file craving. The file craving technique for the most part is to overcome the inherent weaknesses in the traditional data recovery even without any information about the file system. There are many files that can be recovered; images are the most commonly restored files. However, in some cases, JPEG images may be fragmented when stored in the hard disk and it makes it difficult to repair them due to the complexity of determining the fragmentation points. In this research, a technique called Sequential Difference by Segment (SDbS) has been proposed to overcome weaknesses regarding fragmentation detection point of fragmented JPEG images. This technique methodology consists of three steps namely the acquisition of datasets, image segmentation and fragmentation detection. The findings from the third step are the main contributions in this research through the discovery of approaches to detect the fragmentation points using deletion quaternary search and Euclidean Distance to compare pixel values. SDbS has been tested using standard datasets, DFRWS 2006 and DFRWS 2007. They have also been validated using three parameters; average time taken, total image correctly recovered and accuracy of detecting fragmentation point. Based on the results obtained, SDbS has a 19.65% more accurate than SoD, and 0.35% faster than binary search. In other words, SDbS is such a good alternative in addressing problems in detecting the fragmentation points for fragmented JPEG images, compared to SoD and binary search.