Brian's Brain

Jul 15 2008 12:24PM | Permalink | Email this | Comments (0) |
Blog This! using:  Blogger.com | LiveJournal |
Digg This | Slashdot This | add to Del.icio.us

While I'm on the topic of Analog Devices, now's as good a time as any to put cyber-pen to cyber-paper on another long-planned piece. The last two years at the Consumer Electronics Show, I've attended demonstrations of the company's wavelet compression-and-decompression processors, integrated within partners' wireless audio-and-video transmission equipment. The 2007 showcase was fairly underwhelming, with jerky video playback at the receiver end of the link, but a follow-up discussion somewhat clarified the root cause; ADI was (for reasons that remain unclear to me) attempting DVD-derived playback using gear intended for low-to-no frame rate digital signage applications. In (welcome) contrast, this year's version of the demo was pretty pristine, in part a function of the higher transmission bit rate employed.

Why use wavelets, versus a more traditional 'pixel block-based' compression scheme such as MPEG-2, MPEG-4 or VC-1 (aka Windows Media Video)? In part, it's because at a given bit rate threshold, wavelet compression artifacts are often perceived as less egregious to the eye than are DCT (aka Fourier-related)-derived alternatives. On a related note, a wavelet compression-based presentation also tends to degrade more 'gracefully' under inconsistent bit rate conditions (interference, etc). And, ironically, wavelet's proprietary nature is, at least at this early stage in the wireless interconnect game, seen as an advantage by content rights holders and equipment suppliers alike. Hollywood isn't keen to potential pirates being able to tap into the wireless signal in-between transmitter and receiver, and consumer electronics companies are keen to selling you both pieces of gear. A non-commodity protocol neatly accomplishes both camps' objectives.

This resurgence of interest in wavelet compression is, I confess, personally pretty gratifying. A decade ago, I devoted an EDN feature article to unconventional image compression schemes such as wavelet and fractal (a topic I revisited four years later). Wavelet technology subsequently appeared in the JPEG 2000 standard, which never achieved widespread popularity due to traditional JPEG's already large installed base and a perceived low return on the evolutionary infrastructure revamp investment (analogies to MP3 versus AAC or WMA, or to MPEG-2 versus MPEG-4 or VC-1, are apt). And it's also at the heart of the Digital Cinema standard which, 'sexiness' aside, doesn't have a tremendous amount of silicon shipment volume associated with it either at the encoder or playback (theater) sides of the equation. So I suspect ADI's pretty excited about the business potential associated with wireless interconnect in the home, too.