Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

Video Forensic

No description
by

Leo Pang

on 5 December 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Video Forensic

Video Forensics
Noise-based Source Identification
There is a kind of noise generated by the sensor of video source called sensor pattern noise. It is unique depending on different sources.
Video Forgery Detecting Methods
Video Forgery
Video Forgery is a technique for generating fake video by video editing, or altering, combining, or creating new video contents.
Video Forgery
Source Identification
Source Identification
Video Forgery Detecting Methods
Introduction
1. Video Forgery
I will give a counter conception about it and analyze the difficulties of it. So we know why we can reach the target of video forensics.
2.Source Identification
Several methods will be mentioned and briefly explained. In these methods, Identification by noise is the most important one in my presentation.
3.Video Forgery Detecting Methods.
I will emphatically give an explanation about several methods.
Source Identification
Method 1: By defective pixels
Method 2: By quantization characteristics
Method 3: By noise

After an introduction I will talk about the limitation of every method.
Video Forgery Detecting Methods
Method 1: By detecting double quantization
Method 2: Blind detection method for inpainting
Method 3: By ghost shadow

After an introduction I will talk about the limitation of every method.
Some personal opinion about prospective direction
Time for questions
At the begin of the topic, I think we should figure out two problems.
What is Video Forensics? and Why do we need it?
Here is an identification about the Video Forensics:
Methods verifying whether a multimedia content, which can be download from the internet, acquired by a video surveillance system, or received by a digital TV broadcaster, is original or not.
Why do we need Video Forensics?
Let's look at this video firstly...
In my eyes, forged video always appears with two targets:
Maybe for showing off, or for badly illegal benefits.
It's not important what a video forgery for. It transfers wrong information to us and leads to a bad results.
Why we need video forensics is because we want to know whether a video is really shot or manufactured. So we can decide the information sent by a piece of video is right or not.
wrong information
Counter of my Presentation
Details
How can a video be forged?
1. Video editing.
This kind of methods has the advantage when generating a motion object in a video. It looks smoothly, but easy for detecting by data analysis. Cutting, fading in/out, dissolving, wiping are some common ways.
2. Video inpainting.
This kind of methods removes an object from a video without easily being detected. An algorithm is used into the forged process and a frame or part of a frame is altered by computing the motion or repaired by other real part of this video. So it makes the whole image and video looks smooth and without flick when an object is removed.
The difficulties of video forgery
1. A precise object tracking
2. How to remove a object without leaving "ghost shadow"
3. How to change a motion of object smoothly.
4. When there is a moving background, How to change it.

So, because of these challenges, we can track the shortages of a faked video. It provides us ways!
Here is an example of video inpainting
A simple method
Using a dead pixel or a hot pixel to identify a video source
An upgrade method
Using the quantization table
dead pixel
hot pixel
Bo Pang
Instructor: Professor Pradeep Atrey

N=F-Fds
Detecting Forgery by Double Quantization
This kind of method generated from a simple idea that recording an MPEG video, editing it, and re-saving it as another video needs two times of quantization. As a consequence, the image histogram will have some characters.
Blind Detection Method for Video Inpainting Forgery
Let's focus on the histogram
Limitations
This method is very limited by the detected video.
1. The first quantization level is required smaller than the second time.(Q1<Q2)
2. This method only can give a possibility of a forgery not a reliable judgment.
Detecting Method by Ghost Shadow
Due to the temporal discontinuity of the inpainted area, flickers can be produced, which will lead to a visual annoyance, usually called as “ghost shadow”.
Personal opinions about prospective development direction
1. With the significant increasing of wireless equipments, Video Forensic faces the challenge of lossy compression.
2. Because Video Forgery is under self-improvement. Detecting methods only locating themselves on video element is not enough. Mixing with other multimedia components (audio for instance) are necessary.
Z. Geradts, J. Bijhold, M. Kieft, K. Kurosawa, N. Saitoh, “Methods for identification of images acquired with digital cameras,”
Sreelekshmi Das, Gopu Darsan, Shreyas L, Divya Devan
"Blind Detection Method for Video Inpainting Forgery"
Shaxun Chen, Amit Pande, Kai Zeng, Prasant Mohapatra
Weihong Wang, Hany Farid
"Exposing Digital Forgeries in Video by
Detecting Double Quantization"
Full transcript