Comparative study of feature extraction using several wavelet transforms for source camera identification

Abstract

With the advancement in various digital imaging devices, digital images have become ubiquitous today. These can be captured or created by a various digital imaging devices. In many cases it is important to be able to determine the source of a digital image such as for criminal and forensic investigation. This paper presents a novel method to identify source camera by using wavelet transform based feature extraction algorithms. Purpose of this proposed approach is to find best feature extraction component from different wavelet transforms. It consists of seven different wavelet transforms like haar, daubechies, contour let, ridge let, tetro let, curvelet and ripplet. A model of support vector machine is sequentially created and trained by using these features to identify the source camera. In this work, six different source camera models are used for the experiments. The experimental results give high identification rate which is significantly superior to the existing known method. The testing performance justifies the proposed identification method is very useful for source camera identification.