 Overview
Liquid crystal displays (LCDs) have significant advantages over cathode-ray tube (CRT) displays. For example, LCDs generally occupy a smaller foot space and consume less power than CRTs. LCDs, moreover, can provide a higher pixel-density (e.g. 200 pixels per inch) and a higher resolution (e.g. 2560x2048 pixels at QSXGA) on the display screens than CRTs. The current PC system, however, assumes conventional CRT displays and is limited to fully utilize these advantages of LCDs. Our group focuses on various system issues to exploit LCD devices in PC systems.
 Research items
- Digital link technology for high information content display
monitors
The number of pixels and the color depth of PC displays have increased significantly in this decade. Most of the current PCs can display about 1M pixels (e.g. XGA, SXGA) and 24-bit color per pixel. Due to recent revolutionary progress of LCD technologies, LCDs have become able to provide even higher information content - a higher pixel-density and a higher resolution - than conventional CRTs. We believe that such high information content displays improve productivity for conventional PC applications (e.g. business, engineering) and also enable new applications that CRT displays cannot be used.
One of the issues that limit the wide use of high information content displays is the PC-monitor interface; the conventional monitor interface is not adequate for high information content displays because of signal distortions and huge bandwidth requirements. First, the conventional interface uses an analog transmission technique that deteriorates the video signal. The signal distortion has not been a problem because CRTs can display only blurred pixels and because the display resolution, hence the bandwidth requirement of the transmission, has been limited. The signal distortion, however, degrades the image quality to a noticeable level for LCDs because LCDs can display crisper and higher resolution images than CRTs. In fact, several digital transmission techniques for the PC-monitor interface have been proposed to address this issue. The second problem of the conventional interface is the huge bandwidth requirement. Because the interface is based on a raster scan, the PC needs to send all pixel data on the screen for every certain interval. Thus, the transmission bandwidth is tightly coupled with the screen resolution. The raster scan interface, for example, requires 8 Gbps for QSXGA with 24-bit color at 60-Hz refresh rate. None of the currently proposed PC-monitor interfaces can support this bandwidth with a single link.
Our scope includes not only video transmission schemes but also a graphics and monitor architecture that enables high information content displays from a system perspective. This is because we can introduce sophisticated digital techniques to the video interface once the interface is digitized. We are designing a new video system to solve the problems of the conventional monitor interface as well as to provide new functions on the video interface. We use an existing physical layer of a digital video transmission scheme and introduce an upper layer protocol on top of it. The upper layer protocol enables a delta transmission scheme in which the PC sends only modified portions of the screen image to the monitor in an efficient fashion. Thus, the video transmission rate is decoupled with the resolution of the display. The upper layer protocol also supports cascaded multiple monitors that allow us to connect multiple monitors to a single video card in a cascaded fashion. We are also considering to ship a part of drawing functions from the PC video system to the monitor to balance the graphics workload between PCs and monitors.
- Resolution conversion technology of digital images
Imagine what happens if an ordinary XGA video output is displayed on
high content LCD monitors with UXGA or QSXGA size pixels. The video
image would be shrunk to the screen corner if monitors just display
images through XGA pixel numbers. In such case, the readability of
fonts on monitors is low. The best solution to get around this
difficulty is that PCs always output video images consisting of the
exact pixel numbers equipped on monitors. Such solution, however,
cannot be realistic since the life-cycles of PCs are different from
those of display monitors. Therefore, high content LCD monitors must
be built to accept the various resolution video signals from PCs.
As you can easily examine with the above figure, the size
ratios between definitions of display standards are not integer;
these are rational numbers. When a UXGA monitor receives an XGA video
output, for example, it must increase the number of pixels at the
ratio of 25/16 for each horizontal and vertical direction to
represent the received image on the full screen. Actually,
conventional image processing algorithms have troublesomes in image
expansion at ratios of fairly small rational numbers such as 25/16 or
5/4.
This is because conventional studies on image
processing have been done using natural images on televisions or
lowpass filtered images when they are captured; they all can be
assumed of the relative continuity on signal changes. But the
computer screen images mainly consist of graphic-elements that have
tremendous discontinuities of colors and brightness. In addition, the
LCD monitors require us to consider expansion methods based on a
completely new spatial frequency characteristics at the representing
stage, although the CRT monitors which have been used in the
conventional image processing studies represent images like lowpass
filterd ones. Along with this consideration, we are studying
brand-new methods of image expansion that are appropriate for computer
displays.
At the same time, more effective digital
halftoning would be needed as the resolution increases on LCD
monitors. This is because the conventional halftoning for monitors
might not be effective for human visual systems on high-resolution
LCDs. We are also studying new digital halftoning methods with
resolution conversions to obtain visual-effective display systems.
 Publications
- Moriyoshi Ohara et al., "Digital link: High functional digital
monitor interface," SID 99 Proceedings, May 18, (1999).
- Hiroshi Ishikawa, "Digital link that realizes QXGA," seminar note
at "LCD/PDP International '98," October 28, (1998).
 Related information
|
