loading page

Neurocomputing Template
  • nordiana

Abstract

Sharing digital images through social media is becoming very common amongst internet users. A convenient way to share moments and communicate with people all over the world are among the advantages of various social media platforms available through the internet. Nevertheless, this has caused the number of crimes involving digital images to increase. It is well known that each digital image that passes through online social networks (ONSs) is explicitly modified by Web 2.0 tools. Thus, it is challenging for authorities to probe further, including identifying the digital images’ source camera. Considering this limitation, an alternative method to identify source camera based on texture feature for original and OSNs images is proposed here. This technique uses texture feature characteristics, namely, Gray Level Co-occurrence Matrix (GLCM) and Gray Level Run-length Matrix (GLRLM). Original and OSNs images were tested to determine whether the proposed method is robust for both image types and gives higher accuracy than previous methods. Four types of camera models were used in this research. The results prove that the method tested in this study is accurate with an average accuracy of 97.00% and 99.59% for original and OSNs images, respectively, and is capable to read up to 600 images.