Enhancement of image forgery detection using deep neural network transfer learning based technique

Abstract

The process of modifying the content of images has become easier because to the wide availability of sophisticated digital picture technology and image editing software. One useful method for falsifying identity is copy-move forging. Traditional image processing methods typically look for patterns associated with fraudulent material and limit its use in large-scale data classification. On the other hand, DL models have proven to perform noticeably better than other statistical methods. For this reason, an ideal deep transfer learning based copy move forgery detection (ODTL-CMFD) technique was developed in Phase I. The goal of the ODTL-CMFD technique that is being given is to locate the copied regions once a DL model for target picture classification into the original and forged/tampered versions is derived. The political optimizer (PO) using Mobile Networks (MobileNet) model has been derived for the purpose of performing the feature extraction procedure by producing a set of relevant vectors. Finally, the digital photos have been classified as either original or fabricated using an enhanced bird swarm algorithm (EBSA) with least square support vector machine (LS-SVM) model. The application of the EBSA algorithm aids in appropriately adjusting the MSVM technique s parameters, improving classification performance. A number of simulations have been run against the benchmark MICC-F220, MICC-F2000, and MICC-F600 datasets to guarantee the improved performance of the ODTL-CMFD approach. The ODTL-CMFD strategy outperformed the other techniques in terms of many assessment metrics, as evidenced by the experimental findings. newline

Description

Keywords

Citation

item.page.endorsement

item.page.review

item.page.supplemented

item.page.referenced