SSIMWAVE assumes starring role in video quality
A SMALL BUT SWIFTLY growing University of Waterloo tech startup is shaking up the TV and movie industries with a ground-breaking way of tracking video quality on any screen.
Founded just five years ago, SSIMWAVE Inc. has already made waves in the broadcasting and pay-TV industries by pioneering analytic metrics and methods for measuring and monitoring video signals. These metrics and methods, designed to replicate the human visual system, enable the company's software probes to track a viewer's actual quality-of-experience (QoE), rather than just the quality of the video signals themselves.
Known as SSIM (for structural similarity), the complex video analytics algorithm developed by SSIMWAVE has already become the most widely used video QoE algorithm in the world, generating more than 44,000 academic citations as of July. Even more notably, it has been adopted by at least seven U.S. pay-TV operators, including some of the biggest ones, putting the technology in tens of millions of TV homes.
In addition, SSIM has earned Dr. Zhou Wang (pictured below), SSIMWAVE's chief science officer and one of its three co-founders, a Primetime Emmy Award for Outstanding Achievement in Engineering Development from the Academy of Television Arts and Science in the U.S. Wang, an engineering professor at the University of Waterloo who how holds a research chair there, took the prize home, along with a number of other prestigious awards, for the breakthrough technology in 2015.
"It was a big surprise for me," Wang said in a recent interview. "I didn’t even know there was a technical Emmy Award."
As a result, SSIMWAVE has enjoyed strong growth over its five-year lifespan, particularly over the past year. Operating out of cramped quarters in an office building just off the main Waterloo campus, the privately owned company now has 53 full-time employees, up from just seven a year or so ago. Plans call for expanding to as many as 100 employees over the next year as the firm moves to larger space nearby.
But Wang and his two SSIMWAVE co-founders, Dr. Abdul Rehman and Dr. Kai Zeng, are not exactly resting on their laurels now. Encouraged by the stunning success of the original SSIM technology, they have developed a more advanced version of the analytics algorithm, known as SSIM PLUS. It attempts to go even further than SSIM in delivering a unified, end-to-end framework for measuring the QoE of a video program.
"It [the Emmy award] is nice in that it means that we’ve had some impact on the industry," Wang said, "but compared to what we’re aiming for now, that’s nothing."
The SSIM technology works by aping human perception as much as possible. Working with Waterloo students and o
ther volunteer viewers in a specially designed TV lab in the company's headquarters, Wang's research team develops "objective" QoE monitoring probes. Each QoE monitoring probe is designed to “see” and “behave” like a pair of human eyes, perceiving every pixel as regular viewers would.
"What we do is replace humans with SSIMWAVE probes," Wang said. "They [the probes] all speak the same language and are not as expensive as humans."
Wang's team then assigns a specific quality score to each video stream as it travels down the delivery path. The idea is that if the QoE score slips at any point along the way, engineers can quickly identify the issue within the stream and correct it, sometimes before viewers would even notice there's a problem.
If that all sounds pretty complicated, that's because it is. Wang has been working on the thorny video quality measurement issue for about 20 years, dating back to when he was a young PhD student at the University of Texas in Austin. In a 1998 research paper on video quality assessment, Wang argued that the quality of a video should be determined not by measuring the video signal itself but by measuring how well people can actually see it. That was a pretty radical notion back then when network quality-of-service (QoS) was all the rage as the best way to measure video quality and using more bandwidth was seen as the best solution to the problem.
"In this industry, people have a pretty bad habit. They equate bit-rate with quality." – Dr Zhou Wang, SSIMWAVE
"Bandwidth doesn't equal quality," Wang said. "In this industry, people have a pretty bad habit. They equate bit-rate with quality."
In fact, Wang said, with a solid QoE metric in hand, broadcasters and pay-TV providers can actually reduce the amount of bandwidth required to deliver a superior video experience to viewers. That's because the technology enables them to make more efficient use of the bandwidth they have by striking the right balance between perceptual quality and video compression.
We use these perception metrics to drive all kinds of things within this video compression method so that the final outcome of the compressed video is optimized to have the best compromise between perceptual quality and bandwidth. It allows us to gain huge bandwidth savings over typical video coding standards.
"Overall, you can give people a much better experience," he said. "At the same time, you save bandwidth and save money."
Despite some strong initial discouragement by his peers, who viewed his objective video QoE metric as a hopeless quest, Wang decided to tackle the QoE problem head-on. A native of China and the son of two engineering professors, Wang came to the University of Waterloo in 2007 after his post-doc work and early instructional career in Austin and New York when a friend from college recommended the school as an ideal place for his visual imagery research.
Recently inducted into the Royal Society of Canada, Wang continues to refine the SSIM technology to make it more efficient and effective. For him, it's a never-ending battle against video buffering, freezing, stuttering, blurriness, artifacts, low resolution and other quality issues.
"There are a lot more challenges," he said. "This is not an easy job. But it's quite exciting."
