JPEG 2000 is an image compression standard and coding system. It was created by the Joint Photographic Experts Group committee in 2000 with the intention of superseding their original discrete cosine transform-based JPEG standard (created in 1992) with a newly designed, wavelet-based method.
While there is a modest increase in compression performance of JPEG2000 compared to JPEG, the main advantage offered by JPEG2000 is the significant flexibility of the codestream. The codestream obtained after compression of an image with JPEG2000 is scalable in nature, meaning that it can be decoded in a number of ways; for instance, by truncating the codestream at any point, one may obtain a representation of the image at a lower resolution, or signal-to-noise ratio. By ordering the codestream in various ways, applications can achieve significant performance increases. However, as a consequence of this flexibility, JPEG 2000 requires encoders/decoders that are complex and computationally demanding. Another difference, in comparison with JPEG, is in terms of visual artifacts: JPEG2000 produces ringing artifacts, manifested as blur and rings near edges in the image, while JPEG produces ringing artifacts and 'blocking' artifacts, due to its 8×8 blocks.
The advantages of JPEG 2000 compression technology are a result of this underlying structure. Losses are limited to a single frame or less. This brings tremendous benefits to service providers, especially with the growing demands of HD.
- Highly flexible codestream
- Visually lossless quality with no blocking or tiling
- Exceptionally low latency, ideal for interactive applications
- Ideal compression scheme for video over IP applications
- Lower complexity contributes to lower cost and less power & space consumption
The combination of JPEG2000 compression and IP transport constitutes a powerful, robust platform on which to build our new HD/3G and stereo 3D entertainment future. Sending JPEG2000 video over IP networks enables content providers to preserve the highest image quality for all formats. This allows the broadcaster to not only deliver the content to big HD/3D screens but also repurpose the content to deliver a high-quality internet or mobile video viewing experience
Moving JPEG2000 compressed video over IP transport networks addresses the technical challenges facing broadcasters today and well into the future. To better understand why JPEG2000 transport over IP is a powerful platform for cutting-edge broadcast and live digital cinema applications, we we cab deconstruct the way live signals are relayed from a stadium to consumers screen.
As the football game is played, multiple video cameras are rolling and their signals are usually transported and embellished with graphics by a technical team in a central control room or mobile Outside Broadcast (OB) unit.
From there, these live signals travel from the central editing facility or OB Truck at the stadium back to the broadcast network’s operations center via satellite, legacy Telco networks, or more recently IP transport networks. This part of the broadcast process is known as contribution or “backhaul.
Backhaul of high-resolution contribution video can encompass a wide range of applications such as:
- Sending live video of a breaking news event back to the network that will air it in real-time
- Sending live HD video of a high-profile tournament or entertainment event back to the broadcast or cable network that will air it live
- Sending live stereo 3D video of a sports or entertainment event from the stadium or concert hall to digital cinemas for presentation
- Sending stereo 3D video of a live sports or entertainment event to broadcast or cable networks for distribution
- Sending video between multiple digital film post facilities to support creative collaboration on color grading, video editing, and visual effects work
- Sending digital dailies from the shooting location to multiple sites for review
- Studio-to-studio program exchange
- Sending real-time stereo 3D video for medical applications, education, or training
By the time the broadcasters receive the final product, the content may have gone through multiple encoding and decoding stages as it is transported from the venue, processed at the editing facility and then transported to the play-out center. Once the video is ready to transmit, it is encoded a final time and sent to set top boxes in the home. This part of the broadcast process is known as distribution.
JPEG2000’s vital contribution
While contribution and distribution are integral parts of the live broadcast process, each has its own unique concerns and requirements.
For distribution, the goal is effi cient video compression that squeezes the video signal down to fi t into the constrained bandwidth “pipes” to the home set-top box. Sophisticated, costly encoders at the TV station shrink the video signals down such that simple, low-cost decoders in set-top boxes can decode it for display on the home TV. This drastic video compression sacrifi ces valuable picture information to create these smaller “payloads
In contribution or backhaul, the goal is to preserve the picture quality of valuable media assets throughout post production workfl ows regardless of how they are repurposed. This enables the high-quality fi nal product to be distributed over multiple delivery channels in a multi-screen worldBroadcast system designers have two leading compression
technologies that they can use for video contribution:
- Discrete Cosine Transform (DCT): used in inter-frame compression schemes such as MPEG-2 and H264/AVC
- Wavelet compression: an intra-frame compression method used in JPEG200
To better understand the advantages of JPEG2000 wavelet compression for contribution, we must first look at how the alternative technologies approach the job of video compression.
Discrete Cosine Transform (DCT) is well suited to distribution because it was devised as a means of compressing broadband video to a small bit stream that could fit in narrow broadcast or satellite transmission channels. The MPEG family of video compression schemes can employ other tools, in addition to DCT, such as motion vector coding of image differences and entropy coding. Entropy coding can be compared to zipping a file. The computing power required for DCT processes increases significantly in proportion to the size of the image
To get the job done, DCT takes a lossy approach, discarding picture information it deems expendable. With DCT, picture information is not only lost forever, the video further degrades each time it is compressed and decompressed. In addition, if the bit rate is reduced too much, the block structure of the image becomes quite visible and will potentially destroy the viewing experienc
Rather than compressing each sequential video frame in the stream, DCT treats related sequential video frames as a GOP (or Group of Pictures). It then analyzes the adjacent image and encodes only the differences. That means the static part of an image, such as the background behind a news reporter, will only be encoded once per GOP and thereby result in considerable bit rate reduction.
Each GOP is comprised of I-frames (Intra coded frame), P-frames (Predictive-coded frames), and B-frames (or Bi-directionally predictive frames). For the sake of simplicity, let’s say that I, P, and B frames are essentially a “shorthand representation” that can be reconstituted into the original video.
While I-frames can stand alone, restoring the entire video sequence (or group of pictures) depends upon the interplay between the I, P and B frames. If an error occurs in an I-frame, the problem affects the entire GOP, making it more apparent to viewers. For this reason, predictive compression is prone to coding impairments, transmission loss and compounded picture degradation.
In addition, the analytical, predictive process used in MPEG introduces considerable latency or delay to the video. Delay is problematic in real-time signal transport, especially for stereo 3D, since the left and right eye images and audio elements can get out of synch and consequently, ruin the 3D effect.
Quality - You can’t add it later!
The wavelet compression method underlying JPEG2000 is better suited to contribution because it compresses pictures in a visually or mathematically lossless way.
With JPEG2000, the video quality always remains high - a key contribution concern - even when subjected to multiple compression passes.
As an intra-frame encoding method, JPEG2000 encodes each video frame independently. This approach offers many benefits to live video contribution applications. Unlike DCT, JPEG2000 produces very low latency of less than 1.5 frames for encode or decode.
Since each video frame is a key frame containing all of its own picture information, the transport stream can also be edited with frame accuracy. JPEG2000 can allocate 10-bits or even 12-bits at 4:4:4 quality - compared to MPEG4’s 10-bit/4:2:2 quality - a level in line with the demands of video contribution.
Mathematically lossless: uses reversible integer wavelet filtering to ensure that the compressed data has all the information of uncompressed HD-SDI video and therefore provides video transport of equal quality, but with a typical bandwidth savings of at least 60%.
Visually lossless: employs floating point filtering and quantization techniques to provide greater compression with no perceived loss of video quality. Typically, visually lossless JPEG2000 uses from 120-150 Mb/s, while the backhaul of uncompressed HD-SDI needs a 1.5 G/s pipe.