Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1997-10-14
2001-02-13
Mehta, Bhavesh (Department: 2721)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S304000, C358S426010, C358S525000
Reexamination Certificate
active
06188803
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an image processing device and more particularly relates to an image processing device for enlarging or reducing an image by an arbitrary scale by carrying out interpolation of pixel data utilizing a filter coefficient set, of filter coefficient sets corresponding to each phase at the time of dividing the intervals of pixels of an original image by a prescribed dividing number, that is closest the phase of the pixel to be interpolated.
2. Description of Related Art
The use of Cathode Ray Tubes (hereinafter abbreviated to “CRTs”) in displays such as televisions etc. is prevalent, with the handling of analog image signals compatible with various image methods and the changing of horizontal scanning frequency to attain compatibility when displaying these images being common.
However, when handling digital image signals, image resolutions often differ depending on the broadcast transmission method as in the case of, for example, NTSC (National Television System Committee) or PAL (Phase Alternation Line-rate), with the number of pixels in the horizontal and vertical directions of images digitized by these methods therefore differing with each method of broadcast transmission. As there are many broadcast methods including HDTV (High Definition TeleVision) etc., there are therefore many pixel number (resolution) standards. Because of this, it is therefore necessary for a system carrying out digital processing on image data to be compatible with all of these transmission methods and it is therefore necessary for the number of pixels to be converted by an “interpolation filter”.
Further, as the number of pixels for a displayed image is also fixed at a prescribed number for liquid displays and plasma displays that have also recently become widespread, an interpolation filter for converting the number of pixels of a source image to a number of pixels compatible with these displays is also required.
Next, a description is given of an example of an interpolation filter for converting the number of pixels for an image.
First, a description will be given of the conversion of the enlargement or reducing of the image, and the sampling frequency (number of pixels).
The cases of either enlarging or reducing an image or converting (conversion across image standards of differing resolutions) the sampling frequency (number of pixels) of the image are all realized by carrying out calculations to obtain data for pixels that did not exist in the original image with respect to the position of each pixel of the original image. It is therefore possible to carry out the above two processes by utilizing an interpolation filter for carrying out the same arithmetic operations.
FIG. 1
shows an example of a portion of an original image, with circles in the center of the drawing indicating the position of pixels. This portion includes 8 horizontal pixels and 6 vertical pixels (the number of pixels has been taken to be small for simplicity).
Next, a description is given of the case where a source image is enlarged by the scale of, for example, 10/7. This scale is expressed as a ratio of lengths rather than as a surface area. In the case of the enlargement of the image of
FIG. 1
, the pixel arrangement (i.e. the pixel spacing etc.) is kept the same as in
FIG. 1
so that the displayed image standard does not change. The resulting image after this enlargement is carried out is shown in FIG.
2
. As the scale in this case is 1.429 (=10/7) the length of one side of the image is increased by 1.429 times and the number of pixels is increased by approximately 1.429
2
.
With respect to, for example, the horizontal direction, there are 8 pixels in the horizontal direction in the original image, with this increasing to 11 or 12 (adjusted to be close to 8×10/7=11.429) after enlargement. The positional relationship of each of the pixels corresponding to the same portions of the image in the analogous image after enlargement is therefore different to the positional relationship occurring in the original image. The values for data (expressing luminance and color) for each of the pixels after enlargement are therefore different with those of the original image.
FIG. 3
shows the positional relationship of pixels for the horizontal direction of an original image and an image after enlarging in the case of enlarging an image by a scale of 10/7.
In
FIG. 3
, Ri (i=1, 2, . . . ) on the upper side shows pixel data for the original image and Qi (i=1, 2, . . . ) on the lower side represents interpolation pixel data after enlargement. A pixel corresponding to Ri is then arranged at a spacing that is 10/7 times that of the spacing for the pixel corresponding to Qi.
FIG. 3
only shows the situation for enlargement in the horizontal direction, but the situation is the same for enlargement in the vertical direction, and a description thereof will be omitted.
The values for the data for each of the pixels after enlargement is calculated by interpolation filter operations, i.e. carrying out convolution operations on interpolation coefficients from the values for the pixel data for several peripheral source images in response to the corresponding relationship of the positions of each of the pixels of the source image show n in FIG.
3
.
Next, the case of scaling the sampling frequency by, for example, 10/7 without changing the size of the image is described. Changing the sampling frequency in this way is the equivalent of changing the resolution to a higher image standard by a scale of 10/7, i.e. the number of pixels in the horizontal direction is changed by 10/7. In this case, the source image of
FIG. 1
is changed one-dimensionally to an image having approximately 1.429 times the number of pixels, i.e. an image having 1.429
2
times the surface density, as shown in FIG.
4
.
The corresponding relationship of each of the pixels of FIG.
1
and each of the pixels of FIG.
2
and the corresponding relationship of each of the pixels of FIG.
1
and each of the pixels of
FIG. 4
is the same in both cases, as shown in FIG.
3
. The arithmetic operation for converting to an image standard where there are more pixels is the same as the arithmetic operation for enlarging an ag image
Next, a description is given of the case of reducing the source image of
FIG. 1
by a scale of, for example, 10/13.
As the image standard is not changed in the case of reducing the image, the arrangement of the pixels occurring in the image after reducing, i.e. the pixel spacing etc. is the same as for the source image shown in FIG.
1
.
FIG. 5
shows the source image of
FIG. 1
reduced by a scale of 10/13. In this case, the scaling factor is 0.769 (=10/13). The length of one side of the image is therefore reduced by 0.769 times and the number of pixels comprising the reduced image is reduced by approximately 0.769
2
.
For example, the number of pixels in the horizontal direction of the source image is 8 but after reducing this becomes 6 or 7 (adjusted to be near 8×10/13=6.154). The positional relationship of each of the pixels corresponding to the same portions of the image occurring in the analogous image after reducing is therefore different from the positional relationship of each of the pixels occurring in the original image, with the values for the data (expressing luminance and color) for each of the pixels after reducing therefore being different to those of the original image.
FIG. 6
shows the relationship between the pixels for the horizontal direction in the original image and the image after reducing when the image is reduced to a scale of 10/13.
In
FIG. 6
, Ri (i=1, 2, . . . ) of the upper side represent pixel data of the source image and Qi (i=1, 2, . . . ) of the lower side represent interpolation pixel data for after reducing. The pixels corresponding to Ri are arranged at a pixel spacing that is 10/13 times the spacing of the pixels corresponding to Qi.
FIG. 6
shows just the situation for
Iwase Seiichiro
Kanou Mamoru
Kurokawa Masuyoshi
Nakamura Kenichiro
Kananen Ronald P.
Mehta Bhavesh
Patel Kanji
Rader Fishman & Grauer
Sony Corporation
LandOfFree
Image processing device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image processing device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2615661