Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
2000-07-13
2004-10-26
Boudreau, Leo (Department: 2621)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S199000
Reexamination Certificate
active
06810156
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an image interpolation device for interpolating an image.
2. Description of the Prior Art
Interpolation of an image is required, for example, when an image for interlaced scanning is converted into an image for progressive scanning, when an image is enlarged, and when the resolution of an image is increased. Such interpolation of an image is achieved, for example, by increasing the number of lines constituting the image. Specifically, according to conventionally well-know methods, between every two lines of the original image, an interpolated line is added that is produced either by using the lines of the original image immediately above or below this interpolated line intact, or by calculating the average value between the lines of the original image immediately above and below this interpolated line.
However, the former method makes the outline of a figure, such as a slanted line, that exhibits no correlation in the vertical direction appear rugged. On the other hand, the latter method degrades the resulting image by making it appear blurred. To solve these problems, a method has been proposed in which interpolation of an image is achieved by adding interpolated pixels in such a way that, out of a plurality of pairs of adjacent original pixels (note that, throughout the present specification, “adjacent original pixels” denote those of the pixels of the original image which are arranged in lines adjacent to a given line of interpolated pixels) that are so located as to face each other across a given pixel to be interpolated, that pair which exhibits the highest correlation is used to produce this interpolated pixel. Note that, hereafter, a pixel or line to be interpolated will be referred to simply as an “interpolated pixel or line”.
This method will be described in more detail below with reference to FIG.
11
. In
FIG. 11
, the symbols n and n+1 represent lines of the original image, and the symbol i represents an interpolated line. To find the direction in which the highest correlation is observed among three vertical and diagonal directions H
1
, H
2
, and H
3
with respect to a given interpolated pixel X, the absolute value of the difference between two adjacent original pixels is calculated in each of those directions H
1
, H
2
, and H
3
. Specifically, |A(0)−B(0)|, |A(1)−B(−1)|, and |A(−1)−B(1)| are calculated. Then, assuming that the highest correlation is observed in the direction in which the absolute value of the difference is smallest, the average value of the two adjacent original pixels is calculated in that direction and is determined as the value of the interpolated pixel X. For example, if the highest correlation is observed in the direction H
3
among the above-mentioned three directions H
1
, H
2
, and H
3
, then (A(−1)+B(1))/2 is determined as the value of the interpolated pixel X.
In an image having a fine vertical line E
1
that is one pixel wide and brighter than its surroundings as shown in
FIG. 12A
, a part of the line E
1
, when enlarged, is as shown in FIG.
12
B. In an image having a fine slanted line E
2
that is one pixel wide and dimmer than its surroundings as shown in
FIG. 13A
, a part of the line E
2
, when enlarged, is as shown in FIG.
13
B. In either case, the correlations observed in all of the three directions H
1
, H
2
, and H
3
are equal, and therefore it is impossible to determine the direction in which to effect interpolation. To avoid this, the three directions H
1
, H
2
, and H
3
are given different degrees of priority so that interpolation is effected in the direction given the highest priority.
Note that, throughout the present specification, “effecting interpolation in a particular direction” denotes adding an interpolated pixel that is produced by using the two adjacent original pixels that face each other across the interpolated pixel in that direction.
As a result, for example, when the direction H
1
is given the highest priority, interpolation is effected properly in the case shown in
FIG. 12B
; however, in the case shown in
FIG. 13B
, improper interpolation results, because interpolation is effected in the direction H
1
where it should be effected in the direction H
3
.
That is, quite inconveniently, this interpolation method is prone to interpolation errors that occur as a result of interpolation being effected in a direction in which, although the correlation is high, no edge is found. This recognizably degrades the quality of the interpolated image, in particular in terms of the rendering of fine lines, and thus makes it impossible to obtain satisfactorily high resolution in the interpolated image. Note that the direction in which to effect interpolation may be determined by selecting one from among any number of directions greater than three as specifically described above; this, however, increases interpolation errors and thus makes the problem more serious.
SUMMARY OF THE INVENTION
An object of the present invention is to provide an image interpolation device that produces interpolated images with high resolution.
To achieve the above object, according to one aspect of the present invention, an image interpolation device is provided with: absolute difference value calculating means for calculating the absolute value of the difference between two input pixels for each of a plurality of pairs of input pixels that face each other across a to-be-interpolated pixel; edge candidate detecting means for detecting the direction in which the absolute value of the difference between the two input pixels is smallest among a plurality of directions in which the plurality of pairs of input pixels face each other across the to-be-interpolated pixel; edge candidate verifying means for checking whether or not the direction detected by the edge candidate detecting means is a direction in which an edge is found; interpolated pixel value calculating means for calculating the average value of the two input pixels that face each other across the to-be-interpolated pixel in the direction that is recognized by the edge candidate verifying means as a direction in which an edge is found; and interpolating means for determining the value calculated by the interpolated pixel value calculating means as the value of the to-be-interpolated pixel.
Specifically, the edge candidate verifying means may recognize a direction as a direction in which an edge is found if the individual differences of the two input pixels that face each other across the to-be-interpolated pixel in that direction from the input pixels lying adjacently on the identical sides of those two input pixels bear identical signs.
Alternatively, the edge candidate verifying means may recognize a direction as a direction in which an edge is found if the individual differences of the two input pixels that face each other across the to-be-interpolated pixel in that direction from the input pixels that face those two input pixels across and perpendicularly to the line on which the to-be-interpolated pixel lies bear opposite signs.
Alternatively, the edge candidate verifying means may recognize a direction as a direction in which an edge is found if the average value of the two input pixels that face each other across the to-be-interpolated pixel in that direction is nearly equal to at least one of the values of the two input pixels that face each other across the to-be-interpolated pixel and perpendicularly to the line on which the to-beinterpolated pixel lies.
Alternatively, the edge candidate verifying means may recognize a direction as a direction in which an edge is found if the absolute value of the difference between the two input pixels that face each other across the to-be-interpolated pixel and perpendicularly to the line on which the to-be-interpolated pixel lies is greater than or equal to a predetermined value.
In all of these configurations, whether or not the direction in which the highest corr
Boudreau Leo
Conlin David G.
Daley, Jr. William J.
Dang Duy M.
Edwards & Angell LLP
LandOfFree
Image interpolation 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 interpolation device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image interpolation device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3273938