A Forensic Investigation for Suspects' Digital Evidences Using Image Categorization

In computer crimes, even in general crimes, important evidence or clues are increasingly stored in a variety of electronic media, such as computer or mobile devices. The digital data is easily duplicated and it is difficult to discriminate the original from a copy. Further, the digital data can be easily falsified, changed, or deleted from the original data. Therefore, criminal investigations need high level forensic technologies to get better evidences from digital data in suspectspsila computers. This paper is about forensic analyses for digital evidences including a lot of images like pictures and photos. Usually forensic examiners open to check every image files included at hard disks of suspectspsila computers or memory cards. If they have huge amount of images, it takes too much time to check and analyze them. Therefore we use an image filter applying a learning model to divide them into some categories automatically. Through this way, forensic examiners can check out only related image files and then reduce analyzing time. Since, in advance, forensic examiners make some categories for classifying images and input and learn huge amount of image samples to this image filter, accuracy for classifying image files could be improved.