Television – Format conversion – Field rate type flicker compensating
Reexamination Certificate
1997-12-11
2001-08-28
Eisenzopf, Reinhard J. (Department: 2614)
Television
Format conversion
Field rate type flicker compensating
C348S446000, C348S448000, C348S910000, C382S264000
Reexamination Certificate
active
06281933
ABSTRACT:
The present invention relates generally to scan rate conversion systems and, particularly, to scan rate conversion systems employing flicker filters to reduce the variations in interlaced images generated from progressive scan images.
BACKGROUND OF THE INVENTION
Generally, each frame of a video image is represented by a stream of pixel data corresponding to image values (analog, or digital with 1, 8, 16 or 24 bits per pixel) collected at defined positions along a plurality of horizontal scan lines. For example, a typical VGA computer image is represented by 640 8-bit pixels for each of 480 horizontal scan lines. There are two different ways in which the pixel data for a single video frame can be transmitted and displayed.
Referring to
FIG. 1A
, in a progressive scan display, such as a computer monitor, each line of a frame is displayed in order and the entire frame is displayed in a single field. Referring to
FIG. 1B
, in an interlace scan display, such as a television monitor, each frame is displayed in two fields. Each field consists of half the frame's lines and the lines of respective fields are interlaced. For example, the “odd” field consists of the odd lines L
1
, L
3
, L
5
, etc., and the “even” field consists of the even lines L
2
, L
4
, L
6
, etc. Interlace scan data is typical of NTSC, PAL and other television standards.
Presently, there are a number of products that employ a television-type (i.e., interlace scan) monitor to display progressive scan computer data. Examples include WebTV, which uses a television monitor to display Internet images, and various network computers and low cost computer systems that allow users to employ a television monitor as a computer display. This class of products is now described in reference to FIG.
2
.
Referring to
FIG. 2
, there is shown a block diagram generally representative of prior art computer systems
140
that display progressive scan data
121
on an interlaced video display
190
. Such a computer system
140
includes a processing unit
150
, a memory
160
and a scan converter
180
. The memory
160
, which could be a fast random access memory or a slower hard disk drive, holds an operating system
162
, application programs
164
and data
170
. In the conventional manner, the processing unit
150
executes application programs
164
in the memory
160
under control of the operating system
162
. Non-interlaced data
121
is stored temporarily in the memory
160
as video data
172
, which is converted to the interlaced data
141
by the scan converter
180
. Filter coefficients
174
used in the various filters employed in the scan converter
180
can also be stored in the memory
160
.
The scan converter
180
, which can be implemented in hardware or software, generates the interlaced stream
141
in accordance with the vertical and horizontal resolutions and scan rate of the interlaced video display
190
. Typically, a scan converter
180
includes a flicker filter
182
to ameliorate image flicker, which is a problem that arises when images are converted from progressive scan to interlace scan. A scan converter
180
can also include a line memory
184
that provides fast, limited storage for the image data being processed by the scan rate converter
180
and the flicker filter
182
. The problem solved by the flicker filter
182
is now described in reference to FIG.
3
.
Referring to
FIG. 3
, there is shown one frame
122
of an original image represented by the non-interlaced stream
121
, a closeup view
124
a
of that image, and representations
126
a
,
128
a
of the closeup
124
a
as it would appear on subsequent even and odd fields shown on the interlaced video display
190
. Many of the details and horizontal edges of the original image
122
(e.g., the top and bottom lines of the “E” in the closeup
124
a
) lie along a single horizontal line. In interlaced video, such single-line details are scanned at only half the frame rate, causing annoying flickering on the display
190
. For example, the top line of the “E” disappears in the odd field
126
a
and the bottom line disappears in the even field
128
a
. Image flicker results when the fields
126
a
,
128
a
are displayed sequentially, as they would be on the interlaced display
190
.
In the prior art, single dimensional vertical filters (such as the [
1
,
2
,
1
] filter
182
a
of
FIG. 2
) are used to remove this flicker by blurring the offending detail over three frame lines instead of one. When properly executed, such filters completely remove the observable flickering, but at the cost of greatly reduced vertical image resolution.
Another class of filters tries to take advantage of the fact that, in general, flickering is only noticeable on larger elements of the image. For example, thin horizontal lines on the borders of large block characters or graphics flicker a great deal, but the cross element on a ‘t’, or the dot on an ‘i’ doesn't seem to flicker at all. To take advantage of this effect, adaptive filters
182
b
(
FIG. 2
) have been designed that are enabled only when large elements exist on a single line and are otherwise disabled. Such adaptive filters have not been successful as they must switch between the two modes (enabled and disabled) at the pixel rate, resulting in worse artifacts than the flicker they were designed to reduce. Some types of prior art flicker filters are now described in greater detail.
The fundamental problem that causes flicker is the aliasing of an image along the vertical axis. Aliasing is caused when the sample rate of a data system is insufficient to represent the detail in the data. As stated by Nyquist: a wide sense stationary signal can be completely reproduced without error from a set of samples taken from that signal provided that the signal is strictly band-limited to a known frequency (say, FBW), and the spacing of the samples is uniform at a frequency of at least 2× FBW.
In a typical case of converting progressive scan images to interlace scan images, the original image
122
(
FIG. 3
) comprises 480 lines of video. In interlace format, this image is separated into two distinct fields
126
a
,
128
a
of 240 lines each (FIG.
3
). Assuming that 480 vertical samples are sufficient to meet the Nyquist criteria for the source image, each field of only 240 samples would contain an alias error. If the two fields
126
a
,
128
a
were played back simultaneously, the errors would cancel but, in the interlace format, the fields are played back sequentially. Thus, a distortion in alternate fields results that is seen as flicker.
From a Nyquist perspective, the aliasing errors all exist in the upper half of the vertical frequency spectrum (i.e., aliasing errors occur only for high frequency details, which are details that can only be accurately represented with more than 240 samples). Thus, the prior art includes a class of vertical filters (generally known as half band filters) that are applied to the original picture to remove the upper half of the frequency spectrum and thereby filter out the offending elements and eliminate the flicker. Filters of this class that approximate the Nyquist requirement are very complex often having over 35 taps and large arrays of multipliers. Each tap of such a filter requires a full video line of storage to provide the requisite delay. Another problem is that the application of a Nyquist filter (sometimes called a “Brick Wall” filter) creates some serious distortions of its own, such as ringing and loss of detail. For example, small picture details and sharp edges will disappear along with the flicker. In view of these problems with Nyquist filters, another class of filters has been developed that implement a less stringent “Constant Luminance Principle.”
In contrast to the Nyquist criteria, which is based on a mathematical ideal, the constant luminance principle (CLP) rests on perceptual effects. The CLP states: a sampled video signal will have the same overall luminance or intensity level as the original signal provided the original signal
Chrontel, Inc.
Eisenzopf Reinhard J.
Flehr Hohbach Test Albritton & Herbert LLP
Yenke Brian P.
LandOfFree
Images in interlaced formats: a novel method of scan... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Images in interlaced formats: a novel method of scan..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Images in interlaced formats: a novel method of scan... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2514958