Top of page
Content Area

Researchers

Senior Distinguished Researcher FSeishi Takamura
Senior Distinguished Researcher FSeishi Takamura
Senior Distinguished Researcher
Seishi Takamura

Exploring Video Compression Technology from Two Approaches

I mainly work in the field of video compression technology, researching "entity mining coding" and "evolutionary video coding". "Coding" is the compressing of video to reduce the traffic load during data transmission. Among these coding technologies, the method that pursues the essence of what appears in images is called "entity mining encoding". Instead of faithfully sending out all pixels, information is extracted so that we can realistically feel the state of materials and objects we actually perceive, reproducing images that approach the real thing with less data transmission. The other, "evolutionary video coding", is technology that allows a computer to evolve compression algorithms through trial and error much like living organisms, without human intervention.

I Want to Reach the Truth of Things

As a researcher, my mentality is that I "want to reach the truth of things". I was in computer circles when I was a high school student, and a program that my friend made was able to process things so much faster than mine. I was shocked when I realized that the programs could run so differently by just tweaking the algorithm. Later I learned that the algorithm in question was invented during the ancient Greek times, and that it is still the fastest algorithm in the world. I was surprised that technology doesn't become outdated if it reaches the true core of things. I can't say that original experience was necessarily the driving force in me becoming a researcher, but I can say it is the motivation in my research. As a researcher, I think it is important to work with purpose and motivation, but that doesn't mean you should be calculating and selfish in facing various people and work. Since I really don't know what will act as the trigger in advancing my research, I always try to communicate with others interactively and with an open mind.

'Entity mining coding' technology shows images more realistically than the original image before compression
"Entity mining coding" technology shows images more realistically than the original image before compression

There is a Perfect Balance between Basic Research and Development

Many laboratories have a ratio skewed one way or the other between research and development, but I feel like the Media Intelligence Laboratories (hereafter, MD Labs) have that balance just right. First, even researchers entering the MD Labs that just wish to do basic research are transferred to an operating company, being moved closer to the actual scene. Being able to touch on development of the product or service opens the eyes of many to the fun of development. I believe the MD Labs are perfect for researchers looking to take on various challenges, regardless of the research they performed in university. It is an environment that supports the growth of a thick trunk of expertise, as well as various branches and leaves. Additionally, as a member of the NTT Group, it is easy to have my talk listened to by others in gatherings of academic and external researchers, or to create a network. There are also opportunities to study abroad, and many excellent researchers working with you, making it quite a stimulating workplace.

Tackling the Difficult Path to Standardization

I feel like it's worth working on "things believed to be impossible". When I was studying abroad at Stanford University in the USA, the professor in charge of image compression left quite an impression on me, saying "aim for what seems impossible". With that on my mind, I decided to tackle development for technology that seemed impossible 10 years ago, realizing "entity mining coding", which compresses and transmits images more realistic than the original image, and "evolutionary coding", where a computer can create algorithms on its own.

My next objective is to standardize these technologies, much in the same way as JPEG or MPEG. The road is long and the hurdles are high, but that's exactly why I feel like it is a worthy challenge. Since the amount of data being used by videos is increasing every year, even a 1% improvement in the compression ratio can have a large impact on network traffic. As a researcher, nothing would make me happier than being able to contribute to making peoples' lives even slightly more comfortable in the future.

Senior Distinguished Researcher
Seishi Takamura

Profile
Seishi Takamura received his B.E., M.E., and Ph.D. degrees from the Department of Electronic Engineering, Faculty of Engineering, University of Tokyo in 1991, 1993, and 1996 respectively. The same year he completed his doctoral course, he joined the NTT Human Interface Laboratory. After spending time as a visiting researcher at Stanford University (2005-2006), he became a Senior Research Engineer, Supervisor at the NTT Cyberspace Laboratories (now Media Intelligence Laboratories) in 2008. In 2009, he became a Distinguished Technical Member, and in 2016 a Senior Distinguished Researcher. Dr. Takamura is a Fellow of IEEE.

Senior Distinguished Researcher
Seishi Takamura

Profile
Seishi Takamura received his B.E., M.E., and Ph.D. degrees from the Department of Electronic Engineering, Faculty of Engineering, University of Tokyo in 1991, 1993, and 1996 respectively. The same year he completed his doctoral course, he joined the NTT Human Interface Laboratory. After spending time as a visiting researcher at Stanford University (2005-2006), he became a Senior Research Engineer, Supervisor at the NTT Cyberspace Laboratories (now Media Intelligence Laboratories) in 2008. In 2009, he became a Distinguished Technical Member, and in 2016 a Senior Distinguished Researcher. Dr. Takamura is a Fellow of IEEE.

Researchers

Footer Area

Copyright © 2017 Nippon Telegraph and Telephone Corporation