Image Forgery Detection Using Low Dimensional Texture Feature Vector

The emergence of strong program for digital image editing that allows you to easily change the contents of images without leaving any trace signs of such alterations, therefore the authenticity of a digital image is in significant danger. To authenticate a digital image, many digital image forgery detection (DIFD) techniques have been developed. Machine learning is one of the greatest technologies for addressing the issue and assisting in the development of such a system. In this paper, the proposed DIFD has been separated into three steps. The first step involves image processing, followed by feature vector extraction of two handcrafted features from images. Finally, for training these feature vectors, the Gaussian Discriminant Analysis (GDA) classifying model is utilized to distinguish between authentic and fake images. The experiment findings indicate that in the GDA classifier, the (Tamura) texture feature outperformed the (LBP) texture features, as a result, it is the best texture property for fake detection. A comparison of the suggested approach with other with additional modern methods is displayed using the CASIA V2.0 freely accessible dataset.