Television – Image signal processing circuitry specific to television – Noise or undesired signal reduction
Reexamination Certificate
1999-06-18
2003-06-10
Miller, John (Department: 2614)
Television
Image signal processing circuitry specific to television
Noise or undesired signal reduction
C348S587000, C348S645000, C382S167000, C382S260000
Reexamination Certificate
active
06577352
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to object-based image filtering, and more particularly to a method and apparatus for filtering chrominance signals of an image, in which chrominance signals of a frame are sorted into those associated with an object in the frame and those associated with the background in the frame, so that the sorted chrominance signals are filtered independent of each other, thereby avoiding a color bleeding phenomenon occurring due to a signal interference between the chrominance signals respectively associated with the object and background.
2. Description of the Prior Art
Known video coding schemes include a frame-based coding scheme in which the entire portion of a rectangular frame or picture is encoded, and an object-based coding scheme in which only arbitrary shape regions of a rectangular frame are encoded. For the object-based coding scheme, there are various standards such as MPEG-4 and JPEG2000 proposed by International Standardization Organization/International Electrotechnical Commission Joint Technical Committee 1/Sub Committee 29/Working Group 11 (ISO/IEC JTC1/SC29/WG11).
The object-based coding scheme, which uses shape information, is a method in which particular regions or objects of interest or need by the user are extract from an image sequence in accordance with a segmentation method, encoded, and then reconstructed in a frame in accordance with a predetermined order. In such an object-based coding, shape information is used in order to distinguish extracted objects from one another. Such shape information is information for sorting an image into object regions and non-object regions (namely, background). Using this information, it is possible to achieve a signal processing based on object regions of an image in place of the entire region of the image in both an encoder and a decoder. Such shape information may be described in the form of binary shape information or gray scale shape information. The binary shape information is used to distinguish two objects in an image sequence from each other whereas the gray scale shape information is used to distinguish a plurality of objects in an image sequence from one another. For such an object distinction, the binary shape information has a value selected from two values of, for example, 0 and 1, or 0 and 255. On the other hand, for the same purpose, the gray scale shape information has a value in a certain range of, for example, 0 to 255. Shape information used in an object-based image processing includes luminance shape information associated with luminance components of objects in an image signal and chrominance shape information associated with chrominance components of those objects. In particular, shape information associated with chrominance components of objects in an image signal is extracted by sub-sampling luminance shape information, taking into consideration the sampling rate of the image signal and the relationship between the luminance and chrominance signal components depending on the picture type.
For a video scanning scheme used in image processing, there are mainly two schemes, that is, a progressive scanning scheme and an interlaced scanning scheme. In the progressive scanning scheme, an image corresponding to one frame consists of samples acquired at the same sampling time. On the other hand, in the interlaced scanning scheme, an image corresponding to one frame consists of two field sample images, that is, an upper field sample image and a lower field sample image respectively acquired at different sampling times.
In most conventional video coding and processing schemes, an image is processed by frames. For this reason, there are various problems involved in application of those schemes to the object-based coding and processing. In particular, such problems occur remarkably in association with the boundary of an object having an arbitrary shape because the frame-based processing does not take into consideration this object boundary. One problem is a color bleeding phenomenon. That is, chrominance signals associated with the boundary of an object may bleed after being subjected to an image format conversion using a filtering process. Such a color bleeding phenomenon results in a serious problem occurring in processing of images scanned in an interlaced fashion in technical fields such as a broadcasting field required to provide video of a high picture quality.
It is generally known that the sense of sight of the human being is less sensitive to chrominance signals than to luminance signals. For this reason, techniques for decimating/interpolating chrominance signals are mainly used for the purpose of a transmission for video signals in a reduced data amount.
FIG. 1
schematically illustrates a filtering procedure for decimating or interpolating chrominance signals in a state in which the bandwidth of those chrominance signals is limited to a desired range.
Referring to
FIG. 2
, a conventional filtering method is illustrated which involves multiplying input signals Ci by appropriate filter coefficient Wj, respectively, and then adding the resultant signals together. Generally, filters have characteristics determined by their filter coefficients and relationships among those filter coefficients, respectively. Where a median filtering is conducted, filter coefficients for respective input signals may be rendered to be substantially the same because the median filtering is adapted to derive average characteristics of those input signals. The number of filter coefficients and the number of filter taps may be optionally determined in accordance with the use purpose thereof determined by the user. Now, such a conventional filtering method will be described in detail. Where a filtering process is conducted to produce a filtered chrominance signal A, the filtering result obtained through the filters shown in
FIG. 2
can be expressed by the following expression 1:
A=W
1
×C
i−2
+W
2
×C
i
+W
3
×C
i+1
+W
4
×C
i+1
+W
5
×C
i+2
+W
6
×C
i+3
[Expression 1]
where, “C
i
” represents a chrominance component of an image, namely, a chrominance signal, and “W
j
” represents a filter coefficient.
Where this conventional filtering method is applied, as it is, to the object-based video coding and processing, a color bleeding phenomenon may occur. For example, where signals used to produce a filtered chrominance signal A include those associated with a different image region or object from that of the remaining ones, an interference may occur between those signals different from each other. Due to such an interference, a color bleeding phenomenon occurs which results in a degradation in picture quality. Such a phenomenon becomes severe at the boundaries of objects. Assuming that a pixel to be filtered is associated with an object region in the case of FIG.
2
and that “C
i−1
”, “C
i+1
”, and “C
i+2
” have a value of 255 corresponding to an object value, “C
i−2
”, “C
i
” and “C
i+3
” have a value of 0 corresponding to a background value, and “W
1
”, “W
2
”, “W
3
”, “W
4
”, “W
5
”, and “W
6
” have values of 0.1, 0.2, 0.4, 0.6, 0.3, and 0.2, respectively, the value of a chrominance signal A obtained after the filtering for the pixel is 131.58 (A=131.58=0×0.1+255×0.2+0×0.4+255×0.6+255×0.3+0×0.2). Accordingly, it can be understood that the chrominance signal of the pixel having a value of 255 corresponding to the object value is affected by the chrominance signals of pixels arranged the filtered pixel while having a value of 0 corresponding to the background value, thereby causing the filtered pixel to be degraded in picture quality from the value of 255 to the value of 131.58. As a result, a gray strip is formed around the object. Such a color bleeding phenomenon becomes severe at an increased number of filter taps.
SUMMARY OF THE INVENTI
Kim Hae Kwang
Moon Joo-Hee
Park Cheol Soo
Hyundai Electronics Ind. Co., LTD
Lackenbach & Siegel LLP
Miller John
Nissen J. Harold
Yenke Brian
LandOfFree
Method and apparatus for filtering chrominance signals of image does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for filtering chrominance signals of image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for filtering chrominance signals of image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3134444