Around the world, digital video is being transmitted into the homes of consumers from more sources than
ever before. In addition to receiving video from traditional broadcast networks, cable and satellite
providers, consumers are beginning to obtain digital video content from telcos and other IPTV service
providers, as well as downloading content directly from the Internet. The explosive growth of available
digital video content – and in the number of users with broadband connections – is making video display
quality a key competitive factor for the CE industry.
Survey data confirms that the quality of the TV experience is a top-of-mind issue with consumers. According
to survey results, 60 percent of US consumers say picture quality is the most important factor in a new TV
purchase, with brand second at 11 percent and screen size third in order of importance at 9 percent
(source: Hitachi* America and KRC* nationwide telephone survey of 1,055 US adults, 2007).
Once they experience the variety of digital video content made possible by the Internet, consumers naturally
want to access the content of their choice, at any time and any place of their choosing. People who are
becoming familiar with HDTV also want a quality screen-to-screen experience, so as an industry we need to
optimize video quality for each and every device and screen size, from TVs, to PCs, laptops and handheld
media players. At Intel, we recognize that Quality of Experience (QoE) is becoming an increasingly
important competitive differentiator for our consumer electronics customers.
Intel is committed to developing and delivering high-end video processing technology that optimizes visual
quality for any-to-any video format conversions. These include sub-SD all the way to HD for different use
cases and at any location.
What affects video quality?
In the real world, a lot can impact the visual quality of video, and positive factors must outweigh the
negative factors. Quality impairments can begin from the moment the video is created. As anyone who has
used a personal digital camera with available lighting can testify, the original shot can be affected by
vagaries of lighting, noise and camera motion artifacts.
One of the strengths of digital video – its ease of transmission over a variety of service provider and
home networks – can also be a weakness when it comes to quality. Losses that might have negligible impact
on other forms of digital data are readily noticeable in a video image, where a lost pixel or even a
subtle color shift can be extremely distracting. As it is transmitted, video is subject to losses implicit
in video compression, as well as packet losses and the effects of wireless networks.
Consumers simply want video that looks good, and the human eye is particularly adept at detecting quality
problems. Fortunately there is a lot we can do to correct, enhance and condition video to optimize the
consumer’s viewing experience.
We should note at the outset that the filters and techniques originally developed for improving the quality
of analog video have given way to new correction and enhancement methods specifically designed for digital
video. In contrast with gradual noise contamination and signal attenuation, compression and digital
transmission losses create new types of artifacts which degrade the image in a step-wise manner, in other
words there is an abrupt transition from acceptable to annoying visual quality. With its multiple sizes
and formats, digital video multiplies the quality problem in many dimensions. We optimize digital video
quality by sequencing three processing stages: corrective processing, image enhancement and conditioning.
Corrective processing is aimed at reducing negative aspects that may cause user annoyance. Corrective
processing is performed as a first stage to eliminate the propagation or amplification of undesirable
features in the final picture. This process involves minimizing the negatives and maximizing positive
attributes of the picture.
We minimize the negatives through noise reduction filters designed for Gaussian noise, and compression
artifacts such as deblocking, deringing and mosquito noise.
Enhancement is the process of maximizing the positive attributes such as:
- Picture size and resolution – bigger is better
- Color vividness – enlivens the scenes
- Image sharpness – definition of details
- Contrast – the relative definition of foreground and background
- Quality of motion – insures that the visual quality of moving objects that can be tracked by the eye is uncompromised
Enhancement processing is performed after correction so improved features can be propagated all the way through to the output picture.
Picture enhancement involves three variables:
- Sharpness, through both luminance and chrominance transient improvement (LTI/CTI)
- Color quality, including accurate and colorful skin tones, greens and blues
- Adaptive contrast enhancement designed to improve visibility of scene contents at different levels of detail and illumination
Example 1. Sharpness enhancement.
Example 2. Color enhancement.
Example 3. Contrast enhancement.
Conditioning is aimed at matching the specific display size and type, and thus it is performed towards
the end in the sequence. Under ideal circumstances, conditioning does not introduce degradation, or
include methods to compensate for it. Similar to the balance that must be achieved between correction
and enhancement (over-correcting video may compromise its enhancement potential), enhancement and
conditioning must be balanced. We need to ensure that enhancement is optimal for the final image format
regardless of whether the picture size is enlarged or reduced.
Conditioning includes scaling size and modifying the aspect ratio to match the image to the display
being used, and deinterlacing standard definition video so it can be rendered on progressive display
types such as flat screen TVs and PC monitors.
Example 4.
(left) Interlaced fields displayed one at a time.
(right) Advanced motion-adaptive and motion compensated deinterlacing.
As survey results show, visual quality is the most important factor in the buying decision for many consumers,
and this is why Intel is committed to helping our OEM customers optimize Quality of Experience in their products.
As the Internet and digital media continue their convergence, we believe that visual quality in all viewing modes
will be an increasingly important competitive differentiator for digital TVs, multimedia and entertainment PCs,
digital set top boxes, digital media recorders and mobile media players. We believe that hardware-based video
quality optimization in system-on-a-chip (SoC) media processors is most appropriate for specialized consumer
electronics devices. Software-based optimization will be used in PCs that are geared towards general multimedia
applications and that feature powerful multi-core processors which can execute complex algorithms in real time.
Intel is also very active in the development of visual quality metrics and promoting standards for the industry.
The measurement of visual quality currently has three aspects:
- Visual analysis through expert tests by “golden eye” specialists with multi-year experience in video processing algorithms, product design, product development, and recognized expertise in video quality assessment, optimization, and benchmarking
- Standard subjective tests performed at professional labs following ITU protocols
- Objective visual quality metrics such as those under evaluation for standardization by the Video Quality Experts Group (VQEG), which was formed in 1997 to apply the expertise of its participants toward the development of standardized metrics and tools. Many VQEG participants are active in Study Groups of the International Telecommunication Union (ITU).
We anticipate that the development of metrics and standards will eventually lead to visual quality certification
for OEM products, as well as objective metrics for Quality of Service (QoS) management in IPTV and on-demand services.
In addition to being a major player in technologies for access, content management, media storage, content protection,
and other areas, Intel is fully committed to provide superior visual quality and overall Quality of Experience.
With the convergence of the Internet and digital video, consumers will demand visual quality in new CE products, whatever
the content that is being viewed, or the screen size of the device. Intel believes that corrective processing, picture
enhancement and conditioning guided by QoS objectives will be essential platform capabilities for consumer electronics devices.
In addition to hardware and software-based platform ingredients, Intel is involved in the development of objective visual
quality metrics. Objective measures, and standards for quality measurement, may lead to visual quality certification of
CE products and verifiable QoS mechanisms for service providers. By ensuring quality experiences for consumers, quality
standards will encourage the adoption of new usage models and the consumption of new services.