Scalable Decompression with MotionWavelets

By "scalable decompression" we mean the ability to decompress or play back a compressed video at a resolution lower than the original video resolution. For example if we captured and compressed a video sequence at a resolution of 640x480, scalable decompression would refer to the ability to play back the video not only as a 640x480 video but also as a 320x240 or a 160x120 video.

Wavelet compression and the MotionWavelets file format in particular allow very efficient (fast) scalable decompression. In fact, playing back a 640x480 MotionWavelets compressed video at a reduced resolution of 320x240 is just about as fast as playing back any other 320x240 video (and playback speed is roughly proportional to the number of pixels in the video format). In other words, there’s almost no overhead involved in scaling down the resolution at decompression time. This is a very nice feature of wavelet compression, and one that is missing from Motion JPEG.

Benefits & Applications

For Video Editing
In many cases it is inconvenient to perform editing directly on a large-format video sequence (e.g. 640x480 or 768x576) because the video frames occupy so much screen space, and because decompression of the large frames weighs down the CPU during the previewing process, when the CPU is typically overburdened with many tasks, including calculation of transitions and special effects. Therefore a working resolution of 320x240 is often chosen. However, without scalable decompression, decompressing frames from a 640x480 video sequence to a 320x240 resolution can take as long or longer than decompressing it to full resolution. The scalable decompression offered by the MotionWavelets codec solves this problem by providing decompression to a reduced resolution in a fraction of the time. This makes previewing faster and smoother, and allows the user to capture at whatever resolution they need for their final product, without having to worry about the negative effects of a high capture resolution on the editing and previewing processes.

For Video logging / surveillance
A common application for video compression is in a system that captures and stores digital video (and is the digital equivalent of a VCR but with a lot more flexibility). A popular configuration for these systems is to have the video capture and compression performed by a fast PC with large hard drive capacity, but to also provide a comm link (e.g telephone line or T1) over which a remote client PC can query the capture PC in order to also view the live video signal or perhaps browse through video already captured and stored to the capture PC’s disk.

Because the comm link generally has a pretty low bandwidth, it’s necessary for the capture PC to transmit a video signal of reduced frame rate or spatial resolution or both. Using scalable decompression, the capture PC can, with almost no overhead, create two compressed video streams: one high-resolution stream that it saves to its local hard drive, and a second lower-resolution stream that it transmits over the comm link to a remote client PC.

The second step would be to add support to the Video for Windows/ Windows Driver Model installable codec. Once done the codec will answer YES to any queries from applications seeking to decompress an image to one of the supported reduced resolutions. The application will have to specifically ask for a reduced resolution, our codec doesn’t make those decisions on its own.

1. The compressed file can be converted or "scaled" to create a second compressed file of a lower resolution. The lower resolutions available are those with dimensions that are one-half, one-quarter, one-eighth, etc. the dimensions of the original image. In this case the available resolutions are 320x240, 160x120 or 80x60. The scaling creates a new, smaller compressed file that will decompress to the reduced resolution.

   This type of scaling would be used in a situation where the high resolution and low resolution compressed frames are to be sent to or stored in different places, for example if the high resolution frame is stored locally on hard disk, and the low resolution frame is sent over a comm link to a remote site.

2. The compressed file can be decompressed directly to a lower resolution, rather than full resolution. In this type of scaling, no second compressed file is created. This type of scaling would be used when it is desired to store only a single version of the compressed frame, but to be able to play back the video at different resolutions. The benefit of playing back the video at a reduced resolution would be the much lighter demand it would place on the host CPU.

For both variations, the scaling process is very fast because it does not require any recompression of the compressed image, it merely involves extracting portions of the compressed frame and modifying some header information.

For more information regarding Aware's Video compression software products, contact us today.