(IJACSA) International Journal of Advanced Computer Science and Applications, Vol. 10, No. 8, 2019 264 | Page www.ijacsa.thesai.org Robust Video Content Authentication using Video Binary Pattern and Extreme Learning Machine Mubbashar Sadddique 1 , Khurshid Asghar 2 , Tariq Mehmood 3 , Muhammad Hussain 4 , Zulfiqar Habib 5 Department of Computer Science, COMSATS, University Islamabad (Lahore Campus), Lahore, 54000, Pakistan 1, 5 Department of Computer Science, University of Okara, Okara, 56300, Pakistan 2 Department of Computer Science & Information Technology 3 Superior University, Lahore, 54000, Pakistan Department of Computer Science 4 King Saud University, Riyadh Saudi Arabia Abstract—Recently, due to easy accessibility of smartphones, digital cameras and other video recording devices, a radical enhancement has been experienced in the field of digital video technology. Digital videos have become very vital in court of law and media (print, electronic and social). On the other hand, a widely-spread availability of Video Editing Tools (VETs) have made video tampering very easy. Detection of this tampering is very important, because it may affect the understanding and interpretation of video contents. Existing techniques used for detection of forgery in video contents can be broadly categorized into active and passive. In this research a passive technique for video tampering detection in spatial domain is proposed. The technique comprises of two phases: 1) Extraction of features with proposed Video Binary Pattern (VBP) descriptor, and 2) Extreme Learning Machine (ELM) based classification. Experimental results on different datasets reveal that the proposed technique achieved accuracy 98.47%. Keywords—Video forgery; spatial video forgery; passive forgery detection; Video Binary Pattern (VBP); feature extraction I. INTRODUCTION In these days, digital video making has become very handy with the accessibility of video recording gadgets such as smartphones and digital cameras [2, 1]. These videos are an important part of our daily routine and also an important source of information. Digital videos present some of the most convincing documentary evidence to establish the truthfulness or falsehood of an issue under consideration, which is acceptable both inside and outside the court of law. A few years back, digital videos were considered reliable proof, but a widespread availability as well as accessibility of easy-to-use video editing tools (VETs) such as (Pinnacle Studio 20 Ultimate, Adobe Premier Pro, Lightworks and Cinelerra, etc.) [21, 19], have negated this fact. Even a novice user can alter the contents of digital videos in such a manner that it is not possible to distinguish between the original and forged contents of a video with the naked eye. On one hand, video editing is a very useful and important tool for manipulating video scenes in film industry. On the other hand, it enables to forge video contents to distort the evidences for a court and propaganda on social, print and electronic media Therefore, authenticity of a video is a key issue when it is presented in a court as a prof of a crime [12]. Digital video forgery techniques are categorized into temporal, spatial, and spatio-temporal. In spatial category, digital videos are forged by changing contents within the frame(s) which modifies visual information. The object is taken from one location of a video frame and inserted on another place in the same frame or in other frame after some alteration [17]. This category consists of upscale-crop [15], copy-move [18] and splicing video forgery [5]. Temporal tampering (forgery) is done by removing, duplicating or inserting the number of frames from / in a digital video. Both object and frame level forgery is done in spatio- temporal category. Existing tampering detection techniques in digital videos are divided into active and passive. Active techniques need pre-embedded data such as watermarking, digital signatures, etc., whereas passive techniques do not depend on any pre-embedded information. Passive techniques are also called blind techniques. Fig. 1 shows categorization of video forgery techniques. Various passive techniques are proposed to detect spatial video forgery which are not equally efficient for different datasets, static and moving objects. In this research, a robust video content authentication technique is presented. This paper is structured as follows: Section II explains related work; Section III describes a step-by-step research methodology used for development of proposed technique; datasets and performance evaluation parameters are described in Section IV; and experimental work is presented in Section V. Conclusion and future directions are presented in Section VI in the end of this paper.