Neural Network Modeling of Color Array Filter for Digital Forgery Detection Using Kernel LDA

The technology today is well developed that images can be easily manipulated and tampered with various digital tools that we can no longer rely on these images. Doctored images have become ubiquitous and they are so real that they leave very little evidence of being tampered. There have been many attempts to detect such doctored images. Recent proposals have used linear models to represent interpolations in color array filters in digital cameras to identify the forged images. In practice, a nonlinear interpolation is done to perform this task. In this paper, we propose the idea of using Back propagation Neural network as the nonlinear model to represent this interpolation. The features collected from the model were subjected to dimensionality reduction using K-LDA to formulate the NN, NM and SVM classifier and a reasonable success rate of 61.2% was obtained in identifying the forged image.