A 120-Hz image sensor for SUPER Hi-VISION
 STRL is working on the next generation of broadcast services that will deliver higher quality and a greater sense of presence. At present, we are researching and developing full-spec SUPER Hi-VISION, a TV system with 7680 horizontal x 4320 vertical pixels, a frame rate of 120 Hz (progressive scan), bit depth of 12 bits, and a wide RGB color gamut. We recently fabricated a prototype CMOS image sensor for the full-spec SUPER Hi-VISION camera that has the requisite number of pixels, frame rate, and bit depth. The effective image area of this sensor has a diagonal length of 25 mm, and the unit consumes approximately 2.5 W.
 In this presentation, I will describe our new two-stage cyclic A/D (analog/digital) converter and multi-parallel signal output circuit to achieve high-speed operation with a 120-Hz frame rate and 12-bit resolution in a sensor with approximately 33 megapixels. I will also discuss the configuration, specifications, and operation of the image sensor using these circuits. Furthermore, I will discuss the characteristics of the prototype image sensor, as obtained in an image pick-up experiment, and the system configuration of a three-sensor color camera with which we will evaluate the prototype image sensor.


A plasma display panel for SUPER Hi-VISION
 In the SUPER Hi-VISION system, a large high-definition display is needed to reproduce images conveying a strong sense of presence and material perception. At STRL, we have been developing a plasma display (PDP) that is a self-emitting direct-view display device featuring excellent video display performance with high contrast and a wide viewing angle that is relatively easy to produce at large screen sizes. We have just succeeded in building the world's first full-resolution (7680 x 4320 pixels) PDP prototype with a 145-inch screen and a pixel pitch of 0.417 mm.
 In this presentation, I will introduce the panel driving technology that we developed to reproduce stable SUPER Hi-VISION video output on this screen. When controlling the display brightness, we use a multi-line simultaneous scanning technique whereby the pixels of multiple adjacent lines in the vertical direction are displayed at the same brightness, thereby speeding up the panel scanning driver. Furthermore, by suitably selecting the range of brightness to which this technique is applied and the number of lines that are simultaneously scanned, we can achieve stable panel driving characteristics whereby the reduction of picture quality with increasing speed is suppressed, allowing all the lines of the display to be scanned.


Binaural technology for SUPER Hi-VISION audio
 NHK STRL has been developing a 22.2 multichannel sound system for SUPER Hi-VISION. The system consists of 24 loudspeakers, including two for low frequency effects, arranged around the viewers, and it provides a three-dimensional spatial sound. To allow 22.2 multichannel sound to be reproduced in homes, I have investigated the application of binaural technology. Binaural technology is a tool for sound spatialization, in which a sound image is localized by imitating sound propagation characteristics from a point in space to the left and right ears. The sound propagation characteristics are called head-related transfer functions (HRTFs). The application of this technology makes it possible to reproduce 22.2 multichannel sound with a simpler system such as a headphone or conventional stereo loudspeakers.
 In this presentation, I will briefly outline HRTFs, including how it can be measured and adjusted. Then, as a typical application, I will introduce a binaural headphone processor for 22.2 multichannel sound that is being used for audio monitoring at program productions outside the studio, especially those performed in outdoor broadcasting vans. I will also introduce a design of binaural rendering of 22.2 multichannel sound with a simple pair of frontal loudspeakers and describe the current state of our research and our efforts to put this technology to practical use.


Large-capacity transmission technology for next-generation digital terrestrial broadcasting in SUPER Hi-VISION
 Terrestrial analog TV broadcasting terminated in Japan (expect in three prefectures) on 24 July 2011. (In three prefectures in the Tohoku region, it ended on March 31 2012). The transition to digital TV broadcasting was finally completed. At STRL, we are conducting research and development on large-capacity transmission technology with the aim of delivering SUPER Hi-VISION to households by using the next generation of terrestrial broadcasting. On the basis of ISDB-T, which is the international digital terrestrial broadcasting standard developed at STRL, we applied a combination of dual-polarized MIMO and ultra-multilevel OFDM technologies to expand the transmission capacity and built a prototype system that successfully transmitted one SUPER Hi-VISION program over terrestrial waves on two UHF channels.
 In this presentation, I will present large-capacity transmission technologies for the next generation of terrestrial broadcasting. First, I will describe dual-polarized MIMO technology, which enables two different signals to be simultaneously transmitted using horizontally and vertically polarized waves, and ultra-multilevel OFDM technology that expands the modulation scheme up to 4096QAM. I will then discuss techniques such as LDPC coding and inter-polarization interleaving for reducing the degradation in transmission characteristics due to multipath distortion and differences in the field strength between polarizations. I will also describe our field trials in urban areas and SUPER Hi-VISION transmission tests using two UHF channels, which visitors will be able to see at this year's Open House.


Analysis of user behavior in social TV system - teleda -
 Social networking services (SNSs) have become hugely popular, and people around the world are actively using them to share their experiences relating to TV programs they have watched. At STRL, we are developing a social TV system called “teleda” that combines a broadcast service with an SNS so that viewers can enjoy TV programs in a new way. The teleda system includes functions that allow users to express their feelings and share their experiences while watching broadcasts currently on air or previously aired programs. Since this system makes horizontal connections among viewers in addition to vertical connections between the broadcaster and viewers, the viewers can get many opportunities to encounter programs which they’ve never watched before by communicating with other viewers.
 In this presentation, I will describe the service model and system including the teleda API for providing resources such as programs and social graphs. I will also show an overview of the three-month-long field trials on the Internet conducted in fiscal 2010, where approximately 1,000 users experienced the experimental teleda website, and I will discuss the feasibility of services that combine broadcast services with SNSs. Finally, I will introduce our field trials on teleda with extended functions that were held in fiscal 2011, and I will discuss our future plans.


 NHK is working on expanding its sign language services by targeting hearing-impaired viewers whose mother tongue is sign language, in addition to striving to expand its captioning services for the hard of hearing. To achieve these goals, STRL is studying machine translation technology from Japanese text to sign language CG animation. We recently developed machine translation methods to account for the difference in grammar between Japanese and sign language; the language translation is based on a corpus (a database containing a large set of bilingual texts).
 This presentation will cover our machine translation methods and their application to a prototype system for translating meteorological information in Japanese text to sign language CG animation. By using a parallel corpus of Japanese and sign language, we can perform translations without having to know the grammar of sign language. Also, by dividing the parallel corpus into clause and phrase units, it can be used either directly (example-based translation) or indirectly as learning data (statistical translation) to translate a wide range of Japanese expressions into sign language. The prototype system can produce sign language CG animations from our large-scale dictionary.


2012/5/24 (Thu) - 27 (Sun)
Lectures/Presentation