Television – Image signal processing circuitry specific to television – Transition or edge sharpeners
Reexamination Certificate
2000-11-06
2003-07-29
Miller, John (Department: 2614)
Television
Image signal processing circuitry specific to television
Transition or edge sharpeners
C348S673000, C348S606000, C348S252000, C382S266000, C358S447000, C358S532000
Reexamination Certificate
active
06600518
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the field of video processing, and in particular to processes used to enhance a video image.
2. Description of Related Art
The “sharpness” of an image can be improved by enhancing the high-frequency components of the image; that is, by accentuating edges and other portions of an image so that changes are sharp, rather that gradual. A high-frequency enhancement, however, can cause some changes to be overly accentuated, resulting in a “speckling” of the image and other visually unappealing anomalies.
FIG. 1
illustrates a block diagram of a conventional sharpness enhancement device
100
. The sharpness enhancement device
100
includes a convolution kernel
150
that functions as a high frequency filter to identify the high frequency components of the picture. Conventionally, an image on a television screen is commonly referred to as a picture. The terms image and picture are used interchangeably herein, and are meant to include all or parts of a set of data that can be processed to produce a visual rendering corresponding to the data. The convolution kernel
150
computes a convolution value for each pixel of an image, based on the pixel's visual characteristics compared to the characteristics of neighboring pixels. A convolution kernel that is commonly used to provide this value is:
[
0
-
1
/
4
0
-
1
/
4
+
1
-
1
/
4
0
-
1
/
4
0
]
,
the center of this kernel corresponding to the pixel being processed. In this example, the value of the pixel being processed is multiplied by +1, and the values of the pixels immediately above, below, left, and right of the pixel being processed are each multiplied by −¼, and the sum of these multiplied values is the determined convolution value for the pixel being processed. If, for example, the pixel is located in a region of uniform pixel values, the sum of the pixel value minus a quarter of each of four similar pixel values is zero. That is, no enhancements are made within a region of uniform pixel values. Conversely, if the pixel value is 100, and it is surrounded by pixel values of 40 each; the convolution value is 60 (100−10−10−10−10). That is, the larger the change of a pixel's value, relative to neighboring pixels, the larger the convolution value. This convolution value, C, is appropriately scaled by a gain factor, g, at 170, and added to the original pixel value, Yin, at 180, to form a sharpness enhanced pixel value:
Yout=Yin+
g*C.
(1)
The determination of an appropriate gain factor, g, in this example includes four processes
110
-
140
; less costly systems may use fewer processes, with a corresponding lesser quality determination of the appropriate gain factor; additional processes may also be used. The contrast control
110
determines a maximum gain g
1
that can be used without introducing contrast anomalies. That is, if the convolution value C is positive, and the Yin values correspond to a relatively dark area (low Yin values), a large enhancement could produce high-contrasting white values (high Yout values) which will appear as a white sparkles. Similarly, if the convolution value C is negative, and the Yin values correspond to a light area (high Yin values), a large negative enhancement could appear as black speckles (low Yout values). For example, a commonly used range of pixel values is 0 to 255, thereby allowing the pixel value to be processed as a byte, and a commonly used maximum gain factor g
1
in a contrast control
110
is:
g
1=
Yin
/255 if
C
>0 (2a)
g
1=(255−Yin)/255 if
C
<=0 (2b).
The dynamic range control 120 determines a similar maximum gain factor g
2
, to suppress exaggerated overshoots, and the adaptive coring
130
determines a maximum gain factor g
3
, to enhance noise reduction. For example, in a high noise environment, the maximum gain g
3
is kept low if the convolution value C is small, to prevent noise induced changes from being accentuated, while allowing large changes, corresponding to edges in the image, to receive a larger gain.
The output Yout from the adder
180
will be clipped to lie within the minimum and maximum range of pixel values. The clipping prevention element
140
determines a maximum gain factor g
4
to minimize the aliasing that is produced by excessive output clipping. In an example embodiment, the image is divided into blocks, and the number of clippings that occur within each block is used to determine a maximum gain factor associated with each block that will have the effect of reducing the number of clippings within the block. This block-level information is transferred to the pixels by computing a gain factor g
4
for each pixel, based on a bilinear interpolation of the block gain factors. To avoid rapid changes in the time domain, the block gain factors are low pass filtered, using for example a weighted sum of prior gains.
Each of these maximum gain factors g
1
, g
2
, g
3
, g
4
are determined substantially heuristically, each based on a particular set of criteria, and often produce substantially different results. For example, a large Yin value and positive convolution value C will produce a relatively high gain g
1
(equation 2a), but, a large Yin value will often result in a relatively small gain g
4
, to minimize clipping. These potentially conflicting gains are reconciled by a conservative selection process: the gain selector
160
selects the minimum of the gains g
1
, g
2
, g
3
, and g
4
as the appropriate gain to be used. That is, the minimum gain among the maximum determined gains g
1
, g
2
, g
3
, g
4
is selected in order to provide a maximum sharpness enhancement while attempting to avoid the possible effects of over-enhancing the pixel values.
The above described prior art processes are computationally complex, and typically require the use of data derived from a prior image to determine the gains to be used on a current image. The results of these processes can produce anomalous results if fast motion is present between one image and the next. Also, the prior art processes require communication between the processes that can lead to a high bandwidth requirement, particularly when some processes are implemented in software and other processes are implemented in hardware. In particular, the convolution kernel
150
preferably resides in hardware, because of the uniform and repetitive nature of the convolution algorithm, whereas the clipping prevention module
140
preferably resides in software, because of the heuristic rules-based algorithm typically employed. Often, the preferred partitioning of processes cannot be achieved, due to bandwidth limitations. In a multimedia processing system, for example, a preferred partitioning based on the appropriate embodiment to effect each particular task, or to facilitate parallel processing, often cannot be realized, due to the processor-bus bandwidth requirements that such a partitioning would produce. Often, tasks which should be performed in software are embodied in hardware, and vice versa, so as to minimize the amount of data that is transferred via the processor-bus.
BRIEF SUMMARY OF THE INVENTION
It is an object of this invention to provide a less complex method of clipping prevention during picture sharpness enhancement. It is a further object of this invention to reduce the inter-process dependencies during picture sharpness enhancement. It is a further object of this invention to reduce the conflicting results of contrast control and clipping control during picture sharpness enhancement. It is a further object of this invention to minimize the temporal data dependencies during picture sharpness enhancement. It is a further object of this invention to dynamically adjust the picture sharpness enhancement based on the overall enhancement potential of each picture and the overall noise associated with the channel.
These objects and others are achieved by providing a contrast control and clipping device that
Bakhmutsky Michael
Janssen Johan G.
Jaspers Egbert G. T.
Desir Jean W.
Gathman Laurie E.
Koninklijke Philips Electronics , N.V.
Miller John
LandOfFree
Adaptive clipping prevention for picture sharpness enhancement does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Adaptive clipping prevention for picture sharpness enhancement, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Adaptive clipping prevention for picture sharpness enhancement will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3023298