Bandwidth bottleneck broken?
After four years of development, Pulsent Corporation has revealed a new approach to video compression that, it claims, finally breaks the technical and practical barriers to delivering true broadcast quality video over broadband networks.
Pulsent says that its new technology provides a 400% improvement in bandwidth and storage efficiency over existing video compression schemes like MPEG-2.
The breakthrough, it says, will enable telephone companies and other service providers to deliver full-screen, broadcast-quality video into the home at 1.1 Mbps - with the same clarity and resolution that television viewers have come to expect.
Today's leading video technologies, such as MPEG, use a 'block-based' approach to compressing digital video information. This method divides a video frame into arbitrary blocks of a given size, then removes redundancies by attempting to match and reuse blocks from previous frames for the current frame.
When first explored in the 1970's and 80's, this approach was a significant advancement in compression - enabling the storage and transmission of digital television signals at 4 Mbps.
Since then, investigations by academia and industry have resulted in only limited, incremental improvements in compression.
The efficiency of block-based video compression relies on its ability to predict the next frame using blocks, a method known as block-based 'motion compensation.' Over the years, refinements in motion compensation and correction of remaining errors, called residue coding, have played a major role in improving the performance of block-based compression schemes. However, these block-based approaches have inherent limitations since arbitrary blocks bear no resemblance to real objects in natural image sequences.
Pulsent's approach to video compression side-steps the constraints of block-based methods by processing video images in a fundamentally different way.
Instead of dividing the image into arbitrary blocks like MPEG, Pulsent segments the picture into 'objects' based on the visual specifics of the image. After the objects are extracted from the image, their motion is efficiently and accurately modelled using Pulsent's patent-pending predictive techniques.
Because Pulsent tracks each object as a single entity, movement from frame to frame can be predicted more accurately than with other methods. Efficient modelling of common movement patterns such as size changes, object rotation, occlusions, lighting changes and fades results in the modelling of the next frame at very low bit cost. As a result, modelled objects match previous frame objects more often than do block-based approaches.
While Pulsent's compression techniques are unique within the industry, its video payloads are designed for compatible transmission through standard MPEG Transport Stream protocols. Thus, Pulsent video can be delivered through existing network and satellite infrastructures currently supporting MPEG video streams.
Pulsent products will also support side-by-side handling of other codecs such as MPEG, ensuring compatibility with existing applications
The company also anticipates the use of its compression to efficiently 'transcode' video between MPEG and Pulsent formats. For example, a Pulsent transcoder at the head-end of a video delivery network could receive programming available in MPEG-2 and transcode it into Pulsent format to insure very efficient video transmission through the network infrastructure, including the last-mile.