Abstract: This research proposes a novel image splicing detection and localization approach based on the deep convolutional neural network (CNN) learned local feature descriptor. Presented and used to automatically learn hierarchical representations from the input RGB colour or grayscale test images is a two-branch CNN that functions as an expressive local descriptor. The suggested CNN model's first layer, which is specifically made for picture splicing detection applications, is used to extract expressive and varied residual features while also suppressing the impacts of the image contents. Specifically, an optimised combination of the 30 linear high-pass filters employed in the computation of residual maps in the spatial rich model (SRM) is utilised to initialise and fine-tune the kernels of the first convolutional layer.The advancement of digital splicing technology has significantly impacted the progress of digital photo manipulation. This is particularly evident in industries such as newspaper and magazine publication, as well as companies that rely on the verification of photograph authenticity for their publications. Previously, these businesses faced substantial challenges in pre-publication due to the complexities of digital forensics in image processing. However, with the latest developments, the authentication process can now be swiftly addressed with just a few keystrokes. This review is intended to familiarize the reader with various types of digital image splicing forgeries, focusing on the current trend of passive techniques employed to confirm the authenticity of images before they are published

Keywords: Digital image forensics, image forgery detection, Image authentication, , Image Splicing, Passive techniques. Image splicing detection.

Cite:
Dr. Jayenesh H Desai, "A depth analysis of Image Splicing forgery detection", IJARCCE International Journal of Advanced Research in Computer and Communication Engineering, vol. 13, no. 1, 2024, Crossref https://doi.org/10.17148/IJARCCE.2024.13131.


PDF | DOI: 10.17148/IJARCCE.2024.13131

Open chat
Chat with IJARCCE