A Copy-Move Forgery Detection System Using Deep Learning based CNN model and Approximation Wavelet Coefficient

The extensive usage of digital image editing technologies has made image fraud detection an important area of study, particularly in order to guarantee the validity of visual content in a variety of applications like digital forensics, journalism, and law enforcement.Copy-move forgery is the most align type of forgery since it is simple to carry out and effectively hides changes.This study uses a deep learning-based Convolutional type of Neural Network (CNN) model in conjunction with the Approximation Wavelet Coefficient to propose a reliable forgery detection system based on the copymove idea.The suggested technique makes use of the intricate wavelet coefficients of pictures to identify fine-grained forgery indicators.By efficiently breaking down images into multi-resolution components, the wavelet transformation highlights spatial and frequency domain characteristics that are crucial for identifying areas that have been altered.The CNN model, which is trained to precisely locate and identify forged areas, uses these coefficients as input.Results from experiments show how well the system handles a variety of difficult situations, such as noise, geometric alterations, and occlusions.When compared to conventional and current deep learning techniques, the suggested method obtains greater detection accuracy, demonstrating its potential as a dependable tool for image forgery detection in practical applications.