Welcome to the course on:

DIGITAL VIDEO PROCESSING

CS 6870.
Video Content is Very Subjective in Nature
 Applications:
• Surveillance & Security • Video Synthesis and rendering
• Duplicate Detection • Protect copyright
• Event Detection • Tamper detection

• CBVR • Video from sketch

• Detect Saliency • Home and health care

• Categorize (group) in a large database • Entertainment
• Stabilization and improve quality • Storage (coding) and Retrieval
• Mosaicing or panorama • Transmission and playback
• Re-targetting & DA • Traffic and crowd monitoring
• Identify a moving target • Abnormal behavior
• Trail from a network of cameras • Missing object detection
• Autonomous vehicle guidance • Filtering content
• Growth and interaction in bio- • Training in sports, education,
organisms battlefield, etc.
• •
 Key course
contents:

• Video Acquisition - sampling and Representation.

• Motion analysis.

• Video Object Tracking and segmentation

• Video Filtering, enhancement,

• Video stabilization and super-resolution

• Video coding, representation

• Content based Video retrieval

• Video based Rendering

• Video Quality, content recognition and segmentation,

video deblurring (special video – talks)
Text Books:
1. Handbook of Image and Video processing – Al Bovik (Alan C Bovik),
Academic Press, Second Edition, 2005.

2. Digital Video Processing – A. Murat Tekalp, Prentice Hall, 1995

3. H.264 and MPEG-4 Video Compression: Video Coding for Next

Generation Multimedia – Iain E.G. Richardson, Wiley, 2003

3. Digital Image Sequence Processing, Compression, and Analysis – Todd R.

Reed, CRC Press, 2004.

4. References:
IEEE Transactions of :
- Circuits and Systems for Video technology;
- Multimedia;
- PAMI;

IJCV.

Signal, Image and Video Processing (Springer)

 Digital Video processing tasks for various applications

Frame-Rate Conversion - An algorithm for motion-compensated

interpolation. Important frame-rate conversions include 10 f/s to 20 f/s, 24
f/s to 30 f/s (film to TV), and 50 f/s to 60 f/s (PAL to NTSC).

Super-resolution - An algorithm that uses a number of low-resolution

video frames to create a high resolution still frame. Also an algorithm can
take low-resolution video and create enhanced-resolution video.

Mosaicing - An algorithm that takes a number of low resolution frames of a

scene to generate a single large panoramic of the scene.

Deinterlacing - An algorithm for interlaced to progressive conversion.

Possible applications include display of interlaced TV on a progressive
computer monitor, generation of a high quality frame from an interlaced
video of a still scene (e.g. for printing a picture from TV), and standards-
conversion such as 480-line interlaced to 480-line progressive or 1080-line
interlaced to 720-line progressive (important for Digital TV).

Video Stabilization - Handheld video cameras are often afflicted by

unintentional camera movement (jerks). Stabilization process identifies this
unintentional movement and compensates for it.
Segmentation - An algorithm to perform foreground/background
segmentation of a scene. For example, separate the speaker in a video
conference or news show. e.g. to identify the number of people in a scene or
in a room.

Restoration and Noise Reduction - An algorithm to deblur a (blurred or

out of focus) video signal and reduce other artifacts (e.g. motion picture film
is often afflicted by scratches and salt-and-pepper noise).

Motion Estimation - Different algorithms exist for performing motion

estimation. Algorithms may include different methods for gradient-based
and/or block-based estimation, single-layer versus multi-layer estimation.
Comparison criteria could include closeness to the true motion in the scene,
performance when used for compression and complexity.
Video Watermarking - An algorithm to place an invisible watermark in a
video, where the existence of the watermark can be identified by an
associated watermark detection algorithm. In addition to being invisible
and relatively easy to detect, the watermark should be difficult to
remove. For example, adding noise or encoding/decoding the video should
not destroy the watermark.

Object-Tracking - An algorithm to track an object in a video scene.

Examples including tracking a soccer ball in a soccer game, cars driving
down a street, or people moving in a room. The algorithm may be fully
automatic, or semi-automatic where the user initializes the algorithm by
telling it where the object(s) is (are) in the first frame.

Content-Based Representation and retrieval - An algorithm to identify

and extract key-frames from a video sequence so that those frames may
be used, for example, to provide a summary of the video or to enable
efficient searching of the video.
Also represent motion trajectory, texture, color etc.

Video-based rendering
All video storage formats (encoding) are PCM based.

• CCIR 601 used for broadcast stations

• MPEG-4 good for online distribution of large videos and video

recorded to flash memory Discs:
• Video CD (VCD)
• MPEG-2 used for DVDs and Super-VCDs
• DVD
• MPEG-1 used for video CDs
• HD DVD

• H.261 • Blu-ray Disc

• H.263 • Sony ProDATA

• H.264 also known as MPEG-4 Part 10, or as AVC (advanced

Video Coding)

• Theora standardized but still in development. used for video

over the internet (very low bitrate).
Editing sequence, of video clips:

Avid's software and hardware is almost synonymous with the

professional NLE market, but Apple’s Final Cut Pro, Adobe Premiere,
Sony Vegas and similar programs are also popular.
Many interfaces have been designed specifically to handle the
requirements of uncompressed digital video (at roughly 400 Mbit/s):

• Serial Digital Interface

• FireWire
• High-Definition Multimedia Interface
• Digital Visual Interface
• Unified Display Interface
• DisplayPort
• USB

The interface for carrying MPEG-Transport compressed video:

DVB-ASI (Digital Video Broadcast Asynchronous Serial Interface).

Compressed video is also carried using UDP-IP over Ethernet.

Two approaches exist for this:
- Using RTP as a wrapper for video packets;
- 1-7 MPEG Transport Packets are placed directly in the UDP packet
Streaming Media storage size (in megabytes)
= length (in seconds) * bit rate (in kbit/s) / (8 * 1024)

Real world example:

One hour of video encoded at 300 kbit/s (typical

broadband video during 2005 and it's usually encoded in a
320×240 pixels window size) will be:

(3,600 s * 300,000 bit/s) / (8*1024*1024)

= 128 MB of storage.

If this stream is viewed by 1,000 people at the same time

using a Unicast protocol, you would need:

300 kbit/s * 1,000 = 300,000 kbit/s = 300 Mbit/s of bandwidth

Equivalent to around 125 GiB per hour.

Using a Multicast protocol the server sends out only a

single stream that is common to all users.
That stream would only use 300 kbit/s of serving bandwidth.
Examples of Streaming Media systems (usually compressed):

• Flash media server

• Winamp and W-media

• Quicktime

• Peercast

• Freecast etc.

DVB – S2: Digital Video Broadcasting - Satellite - Second Generation;

DVB-SH: Digital Video Broadcasting - Satellite services to Handhelds;

DVB-T2: Digital Video Broadcasting – Second Generation Terrestrial;

TS is the format used to broadcast high-definition TV;

SONY – PS2, PS3;

 List of digital video broadcast standards

ATSC family (North America)

ATSC (terrestrial/cable)
ATSC-M/H (mobile/handheld) DVB family (Europe)

DVB-S (satellite)
ISDB family (Japan/Brazil) DVB-S2

ISDB-S (satellite) DVB-T (terrestrial)

ISDB-T (terrestrial) DVB-T2
1seg (handheld)
DVB-C (cable)
ISDB-C (cable) DVB-C2
SBTVD (Brazil)
DVB-H (handheld)
DVB-SH (satellite)
 The MPEG standards consist of different Parts. Each part covers a certain
aspect of the whole specification. The standards also specifies Profiles and Levels.
Profiles are intended to define a set of tools that are available, and Levels define the
range of appropriate values for the properties associated with them. MPEG has
standardized the following compression formats and ancillary standards:

MPEG-1: is the first compression standard for audio and video. It was
basically designed to allow moving pictures and sound to be encoded into the
bitrate of a Compact Disc. To meet the low bit requirement, MPEG-1 downsamples
the images, as well as using picture rates of only 24-30 Hz, resulting in a moderate
quality. It includes the popular Layer 3 (MP3) audio compression format.

MPEG-2: Transport, video and audio standards for broadcast-quality

television. MPEG-2 standard was considerably broader in scope and of wider
appeal--supporting interlacing and high definition. MPEG-2 is considered
important because it has been chosen as the compression scheme for over-the-
air digital television ATSC, DVB and ISDB, digital satellite TV services like Dish
Network, digital cable television signals, SVCD, and DVD.

MPEG-3: Developments in standardizing scalable and multi-resolution

compression which would have become MPEG-3 were ready by the time MPEG-2 was
to be standardized; hence, these were incorporated into MPEG-2 and as a result there is
no MPEG-3 standard. MPEG-3 is not to be confused with MP3, which is MPEG-1 Audio
Layer 3.
 MPEG-4: MPEG-4 uses further coding tools with additional complexity to
achieve higher compression factors than MPEG-2. In addition to more efficient
coding of video, MPEG-4 moves closer to computer graphics applications. In
more complex profiles, the MPEG-4 decoder effectively becomes a rendering
processor and the compressed bitstream describes three-dimensional shapes
and surface texture. MPEG-4 also provides Intellectual Property Management and
Protection (IPMP) which provides the facility to use proprietary technologies to
manage and protect content like digital rights management.

Several new higher-efficiency video standards (newer than MPEG-2

Video) are included (an alternative to MPEG-2 Video), notably:

MPEG-4 Part 2 (or Advanced Simple Profile) and

MPEG-4 Part 10 (or Advanced Video Coding or H.264). MPEG-4 Part 10 may be
used on HD DVD and Blu-ray discs, along with VC-1 and MPEG-2.

In addition, the following standards, while not sequential advances to the

video encoding standard as with MPEG-1 through MPEG-4, are referred to by
similar notation:

MPEG-7: A multimedia content description standard.

MPEG-21: MPEG describes this standard as a multimedia framework.

 LIST OF TOPICS AND REFERENCES FOR DVP

Introduction:

The Essential Guide to Video Processing, Bovik

Ch. 1: Video Sampling and Interpolation

Handbook of Image and Video Processing, Bovik:

Ch. 3.1: Basic Linear Filtering with Applications to Image Enhancement
Ch 4.1: Computational Models of Early Human Vision;

Motion Detection and Estimation:

The Essential Guide to Video Processing, Bovik:

Ch. 3: Motion Detection and Estimation

Handbook of Image and Video Processing, Bovik:

Ch. 3.10: Motion Detection and Estimation

Digital Video Processing, Tekalp:

Ch. 5: Optical Flow Methods
Ch. 13: Motion Compensated Filtering
Video Enhancement and Restoration:

The Essential Guide to Video Processing, Bovik:

Ch. 4: Video Enhancement and Restoration
Ch. 14: Video Quality Assessment

Handbook of Image and Video Processing, Bovik:

Ch. 3.11: Video Enhancement and Restoration

Digital Video Processing, Tekalp:

Ch. 15: Restoration
Ch. 17: Super-resolution Video Segmentation:

The Essential Guide to Video Processing, Bovik:

Ch. 6: Video Segmentation

Handbook of Image and Video Processing, Bovik:

Ch. 4.10: Video Segmentation

Digital Video Processing, Tekalp:

Ch. 11: Motion Segmentation
 Applications (Retrieval, Surveillance,
Mosaicing):

The Essential Guide to Video Processing, Bovik:

Ch. 5: Video Stabilization and Mosaicing
Ch. 15: A Unified Framework for Video Indexing,
Summarization, Browsing and Retrieval
Tracking: Ch. 19: Video Surveillance

The Essential Guide to Video Processing,

Bovik:
Ch. 7: Motion Tracking in Video

Handbook of Image and Video Processing,

Bovik:
Ch. 4.11: 2D and 3D Motion Tracking in
Digital Video Optimization:

Digital Video Processing, Tekalp: Digital Video Processing, Tekalp:

Ch. 9: Methods using Point Ch. 7: Pel-Recursive Methods
Correspondences Ch. 8: Bayesian Methods
Ch. 10: Optical Flow and Direct Methods