Developed method for detecting forgery videos by Chroma by tracking residual noise and edge detection
Keywords:
Chroma, forgery, foreground separation, background, edge detection, optical intensity, object detection, frame difference, noise, boundary matchingAbstract
This research presents a new method for detecting Chroma forgery, by discovering the objects that make up the foreground of the video by subtracting the foreground of the video from its background, and then comparing the boundaries of these objects with the boundaries of the remaining noise-free objects from the Chroma etching process, which constitute the intro video, and detected by relying on edge detectors and noise added, and in the event of a match with a certain threshold difference, it can be confirmed whether the video was forgery or not.
We have tested the proposed method on a set of database consisting of 20 fake chroma videos produced by us and 7 videos based on chroma in their production and other fake methods downloaded from the Internet. The experimental results of this method showed an approximate accuracy of 85.2% in the process of detecting forgery, from For different chroma colors and their different techniques, the method also showed the ability to detect other types of forgery such as the very common copy-transfer type, in addition to its ease and great speed in the detection process.