Image analysis – Image enhancement or restoration – Image filter
Reexamination Certificate
1999-08-31
2002-04-30
Johnson, Timothy M. (Department: 2623)
Image analysis
Image enhancement or restoration
Image filter
C375S240290, C382S264000
Reexamination Certificate
active
06381373
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to an image signal processing apparatus, an image signal processing method, and an image signal decoding apparatus. More specifically, the present invention in directed to an image signal processing apparatus, an image signal decoding apparatus capable of reducing a quantizing error when an encoded image signal is decoded to be displayed.
2. Description of the Related Art
In such a system, a moving picture signal is recorded on a recording medium such as a magneto optical-disk and a magnetic tape, this moving picture signal is reproduced so as to be displayed. An moving picture signal is transmitted from a transmitter via a transmission path to a receiver, as in a TV conference system, a TV telephone system, and a broadcasting appliance. In these systems, since the transmission path and the recording medium are utilized at a high efficiency, the image signals are compressed and coded by utilizing the line correlation as well as the frame correlation of the video signals.
A description will now be made of the high-efficiency coding of the moving picture signal.
Conventionally, moving image data such as video signals contains very large quantities of information. Thus, to record/reproduce this moving image data for a long time duration, the recording mediums whose data transfer speeds are very high are necessarily required. As a result, a large-sized magnetic tape and a large-sized optical disk are needed. When moving image data is transmitted via the transmission path, or broadcasted, since an excessive data amount must be transferred, such moving image data could not be directly transmitted via the presently existing transmission path.
Under such a circumstance, when a video signal is recorded on a compact recording medium for a long time, or is used in a communication and a broadcasting system, the video signal must be processed in the high-efficiency coding for a recording purpose, and further such a means for decoding this read signal at a higher efficiency must be employed. To accept such needs, the high-efficiency coding methods by utilizing the video signal correlation have been proposed. As one of these method, so-called “MPEG (moving picture experts group) 1 and 2” have been proposed. This method has been proposed as a standard method in ISO-IEC/JTC1/SC2/WG11 conference. That is, such a hybrid system combining the movement compensation prediction coding with DCT (discrete cosine transform) coding has been employed.
The movement compensation prediction coding corresponds to the method for using correlation of the image signal along the time base direction. The presently entered image is predicted from the signal already being decoded and reproduced, and only the prediction error is transmitted, so that the information amount required in the data coding is compressed.
The DCT coding is such a coding method that signal power is concentrated to a specific frequency component by utilizing the frame two-dimensional correlation owned by the image signal, and only the concentrated/distributed coefficients are coded. As a result, the information amount can be compressed. For instance, the DCT coefficient at a flat picture portion of which self-correlation is high is concentrated and distributed into the low frequency component. Thus, in this case, only the coefficient concentrated/distributed to the low frequency component is coded, so that the information amount can be reduced.
It should be noted that although the MPEG 2 system is employed as the coding device in the specification, many other coding systems except for the MPEG system may be applied.
In general, to improve an image impression in a TV monitor, a filter for emphasizing the image is employed. There are various sorts of emphasizing process filters. For example, there is such a high-pass filter for emphasizing a high frequency component of an image signal. Also, a so-called “contrast filter” is provided to amplify an amplitude of an image signal. There is another filter for converting density gradation.
These emphasizing filters not only emphasize the image signal to increase the visual impression, but also the noise contained in the signal. As a consequence, when the image signal contains many noises, these noises would become apparent, resulting in deterioration of visual impressions.
The output signal to the display unit of the TV monitor, and thus, the above-explained emphasizing process is performed after the digital signal is converted into the analog signal. This is shown in FIG.
1
and FIG.
2
.
FIG. 1
shows such a case that an input to a TV monitor
400
is a digital signal. The digital input signal is D/A-converted into an analog signal by a D/A converter
401
, and then is emphasized by an emphasizing filter
402
. Thereafter, the emphasized signal is supplied to a display unit
403
for representation.
FIG. 2
shows another case that an input to a TV monitor
410
is an analog signal. The analog input signal is emphasized by an emphasizing filter
412
to be supplied to a display unit
413
for representation.
In the case of the analog image signal, random noises such as white noise are major noises. The noises of the digital signal are block distortion, and quantizing noises near edges. These noises are locally produced and own higher correlation-thereof. When the noise contained in the digital image signal is emphasized, the visual impression would be greatly deteriorated, and would give unnatural impression.
Normally, a digital image signal is quantized by 8 bits. In the normal image signal, quantizing noise could not be visually recognized. In other words, no discrimination can be made of a 1-bit interval in 8 bits. However, as shown in
FIG. 3
, when an image signal is simply increased in a flat manner, this quantizing error, namely 1-bit interval can be recognized. This is because a human observation is very sensible to this flat portion, and steps of 1-bit interval are continuous.
A similar phenomenon appears as to an image with a better S/N ratio. When an image owns a low S/N ratio, the 1-bit interval is mixed with noises, which cannot be therefore observed. However, as to an image having low noise, the quantizing error (1-bit interval) can be discriminated. This phenomenon especially occurs in the noise-eliminated image, and the signal and CG produced by the image signal generating apparatus.
When the image signal is coded, a similar phenomenon occurs. A general image signal contains noise. When the coding bit rate is high, this noise component is also coded-to be transmitted. When the coding bite rate is low, this noise component could not be transmitted. At this time, in the MPEG coding system to perform the block processing, this may be observed as a block-shaped noise. If such a block-shaped distortion is continued, even when this corresponds to a step of 1-bit difference, it could be visually recognized. Since this is observed as a pseudo contour, it is called as a “pseudo contour”.
FIG.
4
A and
FIG. 4B
represent such a case that pseudo contours are produced.
FIG. 4A
indicates a two-dimensional pattern displayed on a screen.
FIG. 4B
represents a signal level on a line a to “a” of FIG.
4
A.
A similar phenomenon will occur when a decoded image signal is emphasized. When the image signal is emphasized, a 1-bit difference would be widened. This phenomenon will now be explained with reference to FIG.
5
A and FIG.
5
B.
FIG. 5A
shows a case that a step of a 1-bit difference is converted into an analog signal. When this signal is emphasized, as shown in
FIG. 5B
, the 1-bit difference would be widened. As a result, this 1-bit difference could be visually recognized. This phenomenon is visually recognized as a pseudo contour.
When the image signal is decoded in the above-described manner, the quantizing error can be visually recognized to produce the pseudo contour.
Also, there is another possibility that the quantizing error can be discriminated also in the not-coded image signal. This may be
Suzuki Teruhiko
Yagasaki Yoichi
Frommer William S.
Frommer & Lawrence & Haug LLP
Johnson Timothy M.
Polito Bruno
Sony Corporation
LandOfFree
Apparatus and method for reducing quantization error in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for reducing quantization error in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for reducing quantization error in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2836620