Markov features based DTCWS algorithm for online image forgery detection using ensemble classifier in the pandemic

Abstract

The unexpected blast of the COVID has been found everywhere in the world, and everyone has been shocked unusually. In this critical situation, the number of online activities increases, whether related to online education, research, business meetings, virtual conferences, or virtual court. Under this pandemic situation, digital images are the only source of information that can be generally shared and visualized in virtual conferences and social media, and it's challenging to share the document for forgery detection. Today, it's straight forward to forge these images using image-editing software, and it's essential to detect image forgery for such images. In this paper, an efficient novel Discrete-Time Cosine Wavelet and Spatial (DTCWS) Markov feature-based algorithm has been designed for the detection of such forgery, especially for this pandemic situation. For this work, high-dimensional Markov features have been extracted in the DTCWS domain, and the dimensionality of these Markov features has been reduced with Principal Component Analysis (PCA). Furthermore, the co-occurrence matrix has increased the correlation among coefficients. For classification, an optimized ensemble classifier is used for evaluating the results instead of using a support vector machine classifier. Due to the time constraint in online activities, the proposed algorithm shows the best accuracy of 99.9% without taking too much time and fewer complexes compared to the current work.