Data Hiding for Video-in Video - PowerPoint PPT Presentation

1 / 13
About This Presentation
Title:

Data Hiding for Video-in Video

Description:

The Madonna video is embedded for real time playback along with the host video. The Madonna video is encoded using MPEG at a bit rate of 294 bytes/frame ... – PowerPoint PPT presentation

Number of Views:800
Avg rating:3.0/5.0
Slides: 14
Provided by: pvpo
Category:
Tags: data | hiding | madonna | video

less

Transcript and Presenter's Notes

Title: Data Hiding for Video-in Video


1
Data Hiding for Video-in Video
  • by Mitchell Swanson, Bin Zhu and Ahmed Tewfik

Summarized By Pritam Potnis
2
What did they Suggest ??
  • A scheme for hiding high bit-rate supplementary
    data, such as secondary video, into a digital
    video stream by directly modifying the pixels in
    the video frames.
  • The technique requires no separate channel or bit
    interleaving to transmit the extra information.
  • The data is invisibly embedded using a perception
    based projection and quantization algorithm.
  • The Data Hiding algorithm supports user defined
    levels of accessibility and security.
  • Algorithm is illustrated by using Real Time
    Video-in-Video and Speech-in-Video examples.
  • Also demonstrates the robustness of data hiding
    procedure to video degradation and distortions.

3
Highlights
  • Hidden Data remains intact as the host video is
    compressed, stored and transmitted.
  • The Embedded data can be extracted from the video
    stream automatically or when the user selects a
    particular temporal/spatial region.
  • The embedded data is invisible within the host
    media
  • Hiding data directly into the video data,
    resolves the problems like synchronization,
    during playback and additional storage.
  • Hiding information in spatial or temporal regions
    lets the data remain in these regions if they are
    removed and pasted into another location.

4
Diagram of Video Data Hiding Algorithm
5
Diagram of Block Based Projection
6
Diagram of Embedding Data
a) Quantization of projection by masking
threshold T, and b) Embedding a 0 by
perturbing the quantized value by T/4
7
Lets get to the Specifics
  • The Approach is based on two steps.
  • Linear Projection
  • Quantization and Perturbation
  • Each Frame in the video sequence is broken into
    small blocks and transformed into the DCT Domain.
  • Each Block is projected onto a pseudo random (to
    generate unique author defined signatures) author
    defined direction.
  • The DCT of the pseudo random sequence y is
    computed Zijdct(y)
  • The data is embedded by modifying the projection
    p
  • Projection is quantized according the masking
    based threshold TminMij
  • The new projection p contains the hidden data.
  • The new image is v v (p-p)z
  • To Extract Data the receiver projects v onto z
    (z is computed from x by authorized receivers),
    removes the integer portion of the projection,
    and evaluates the remainder
  • b 1 If lt v, z gt -lt v, z gt gt 0
  • b 0 otherwise
  • The original image block v is not required to
    extract the embedded data
  • One or more bits per block can be inserted by
    projecting the block onto multiple directions.

8
Video-in-Video Application
  • Original Broadcast News Video (b) Broadcast News
    Video with Embedded
  • Madonna Video.
  • The Madonna Video is the Real -Time Video

9
Whats Happening in theVideo-in-Video Application
  • A 311 frame, 120 x 160 grayscale video of Madonna
    ? video is embedded in an equal length sequence
    from another movie.
  • The Madonna video is embedded for real time
    playback along with the host video.
  • The Madonna video is encoded using MPEG at a bit
    rate of 294 bytes/frame
  • The frames of the Broadcast News Video are of
    size 240 x 360
  • No degradations of the video sequences are visible

Unlike PIP, no separate data streams for each
video are transmitted for this procedure. Only
host video stream is transmitted and the embedded
video stream is extracted and displayed in a
window within the host video, streamed to a
second display device
10
Speech-in-Video Application
  • Original Classroom Sequence (b) Classroom
    Sequence with Hidden Data
  • Four Embedded Speech Signals

11
Whats Happening in the Speech-in-Video
Application???
  • Speech is also supported by this algorithm.
  • Four speech streams are embedded and played back
    in real time.
  • Video 250 frames (Size 360 x 240) Original
    Speech Stream 8KHz
  • A Code Excited Linear Predication (CELP) voice
    compression algorithm is applied to each speech
    signal
  • Resultant compressed waveforms are coded at 2400
    bits per second.
  • In this way the Classroom Video can be
    transmitted with four language options for a
    multilingual audience.

12
This Technology
  • Can be used to enhance video watching experience
    by giving viewers the capability to control and
    personalize what they view.
  • Can be used for interactive digital TV or
    Video-in-Video without a separate channel.
  • Can use private author keys, allowing only
    privileged users to access the data.
  • Employs a masking model to guarantee the hidden
    data is invisible and maximizes the strength of
    the hidden data

13
THANK YOU !!!!
Write a Comment
User Comments (0)
About PowerShow.com