Please note that JavaScript and style sheet are used in this website,
Due to unadaptability of the style sheet with the browser used in your computer, pages may not look as original.
Even in such a case, however, the contents can be used safely.

Real World Interfaces

Real World Interface Technologies

To improve the productivity of companies and improve our daily lives by leveraging information and communication technology (ICT), we need to sense the events that occur in the real world in a manner that can be processed by ICT systems (cloud computers), and feed back the processing results to the real world (= creation of interface). The Information and Media Processing Laboratories work on research and development to allow the real world to be handled as multidimensional information such as images. Concretely, our R&D activities are being carried out in the following areas related to multidimensional information: 1) sensing signal quality enhancement, 2) sensing signal analysis, 3) highly efficient transmission of sensing information, and 4) rendering of processing results.

Sensing signal quality enhancement

(A) Image quality enhancement technology

We have developed technology to automatically categorize image scenes and automatically adjust them to fit human’s preferred one using the most suitable techniques , such as sensor noise reduction and memorized color reproduction, for each category.s. These technologies can be completed in real-time even with limited computing resources, and make it possible to always take optimum pictures on mobile phones without users having to be aware of modes and settings.

(B) Hyperspectral signal correction technology

We have developed a technology to restore with high fidelity the spectral power distribution of illumination and the surface reflectance of objects from spectral signals captured with a hyperspectral camera. This technology, which makes it possible to achieve stable object identification using a hyperspectral camera under all kinds of weather conditions, is expected to be applied in various fields such as sea rescues and resource exploration.

Sensing signal analysis

(A) Video content identification technology

We have developed technology to automatically and instantaneously identify moving pictures with a high degree of accuracy by extracting the fingerprint information (video signature) identifying a video scene. This technology, which makes it possible to detect modified video copies made with a video camera or some other means and illegally uploaded to the Internet, can be expected to help achieve more secure content distribution services. This technology has been made into an international standard as ISO/IEC 15938-3/Amd.4, MPEG-7 Video Signature Tools.

(B) High-accuracy & real-time object recognition

We have developed technology that recognizes in real time and with high accuracy objects and instantaneously displays detailed information about them by simply directing the camera embedded on a smartphone or tablet at them. This technology does not require any preparation of visible markers in advance on objects, such as barcodes or IC tags, and therefore, can contribute wide range of applications where existing ID marker technologies are not used.

(C) Semantics of video scene understanding technology (fixed-point camera application)

We have developed a system to assist the accurate understanding of situations at remote offices being observed with fixed-point cameras and microphones. By analyzing the cameras’ video data and the microphones’ audio data and displaying them superimposed as onomatopoeia (imitative/echo words and mimetic words), the area or situation of interest can be grasped at a glance, allowing convenient remote communication.

Sensing information transmission

(A) Ultra-low delay video coding technology

We have developed technology to reduce the delay required for video compression for efficient transmission to levels not perceptible by human senses. Use of this technology for remote control via video and two-way video communication, among other applications, allows stress-free and highly realistic remote communication.

(B) Next-generation video coding technology

High-quality video compression coding methods are widely used as a fundamental technology that supports digital HD and one-segment broadcasting. To implement high-quality HD video transmission in the mobile environment, we are developing high efficiency video coding methods as the next-generation video formats and promoting their adoption as international standards, in a bid to achieve both ultra-low rate and extremely high-quality coding technology.

(C) Lightweight high-compression image coding technology

We have developed still image compression technology that allows high compression efficiency with little computation processing. This technology achieves compression efficiency comparable to that of the latest coding standards such as JPEG 2000 for less than one tenth the processing load. As a result, large amounts of high-quality images can be stored and transmitted even in limited processing performance and bandwidth-constrained environments such as mobile devices. This technology is employed as the main lossless compression method on Venus probe Akatsuki.

Information rendering

(A) Natural interactions technology

We have developed a new user interface technology that projects the needed information dynamically on any preferred surface by understanding the user’s situation. By combining a tiny video projector with a movable camera for 3D shape measurement and recognition, , this technology allows for example the easy-to-use information transmission between devices while presenting suitable feedback information in physical space between the devices. Intuitive interaction can be achieved as a result.

Image quality comparison at 1/60 compression