Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
2000-11-27
2004-04-06
Johnson, Timothy M. (Department: 2625)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S293000, C382S295000, C382S299000, C358S001200, C358S528000, C345S671000, C345S660000, C384S561000, C384S581000
Reexamination Certificate
active
06718072
ABSTRACT:
PRIOR FOREIGN APPLICATION
This application claims priority from Japanese patent application number 11-365016, filed Dec. 22, 1999, which is hereby incorporated herein by reference in its entirety.
TECHNICAL FIELD
The present invention relates to a method and an apparatus for increasing, using a rational number multiplier, the resolution of an image consisting of a set of pixels that are represented by digital values.
BACKGROUND ART
Most conventional image expansion techniques employ a natural image, such as a scenery picture, as an expansion target, and substantially there are no stepped edges, such as thin lines, in an image. This is mainly because information is recorded, while the stepped edge is transformed, by the low-pass effects of an image pickup apparatus, into a form represented by, for example, a sigmoid function (a differentiable, continuous function, f(x)=1/(1+(e
−x
)). Therefore, conventionally, improvements have been performed to determine how to make the outline, which tends to be more ambiguous than necessary, to appear natural, with the assumption that the original image was obtained by sampling a low-pass filtered image.
For resolution conversions, which require the least computations and can be easily implemented, the replica and nearest neighbor methods are used. With the replica method, the simplest, an image is expanded (n+1)
times by copying the same pixel at every n-th pixel. And with the nearest neighbor method, an image is expanded by copying the pixel of the original image that is closest to the coordinates obtained by converting the resolution. Both the replica method and the nearest neighbor method provide substantially the same results; that is, the mixing of colors among pixels does not occur, and changes in tones are persistent.
Also, a bi-linear interpolation method has been established for the conversion of resolutions. With this method, the coordinates of a pixel point for a resultant image are inversely projected onto the coordinates of the original image. Then, the pixels (four surrounding points, two points on each of two sides, or only one point located at the identical coordinates) of the original image in the vicinity of the coordinates are weighted, using distance, to obtain an average value, which subsequently is regarded as the color of the resultant image.
Furthermore, a bi-cubic interpolation method is also well known. This method applies the same principle as does the bi-linear method, but for two surroundings in the vicinity (16 points). According to this method, it is assumed that the differential continuity and the value change (gradient) of the original image are sufficiently moderate. Colors, which are enhanced by the weighing of parameters, are more clearly defined than they are with the bi-linear method.
In addition, there is a multirate system that is obtained by generalizing the preceding interpolation methods and that requires more computations. The basic configuration of the multirate system provides for low-pass filtering to be performed after up-sampling using zero-value interpolation, and down-sampling is performed to expand an image at a predetermined rate. This configuration can theoretically include the bi-linear method and the bi-cubic method by adequately selecting the low-pass filter. Actually, in many cases a poly-phase configuration or a filter bank configuration is mounted in order to increase the computational efficiency.
However, according to the replica method and the nearest neighbor method, the pixel of an original image is simply held at a new spatial sampling point, and the expanded line width differs depending on the coordinate positions. Since the sensitivity to frequency components of human eyes is high to a low frequency for the angle unit, a serious problem occurs in the readability of a line width in these methods. The study of an eye model is described in reference IEEE Trans. on Image Processing, vol. 4, pp. 1460-1464, 1995. According to this reference, when an oblique line is expanded, it can be assumed that the distortion in a low frequency band, such as an increase in the width of jaggies, can be easily perceived and that the quality will be deteriorated.
Further, according to the bi-linear interpolation method, when the coordinates that are determined for an inversely projected image are located at the same distance on either side of a one-dot width line, the resultant image will invariably represent a line as wide as two lines of a half a color. Therefore, problems occur relative to the uniformity of colors, the legibility of characters and the reproduction fidelity of colors. Although an image in a photo may appear to be satisfactory, the overall impression is of a blur.
Furthermore, the bi-cubic interpolation method has the same problem as does bi-linear interpolation method, where one line is changed into two lines half a color. Thus, the accurate reproductivity of colors of the screen image of a personal computer (hereinafter referred to as a PC) is still problematic. Also, there is a slight occurrence of ringing at sharp boundaries between middle tone colors.
Moreover, when an arbitrary low-pass filter is employed for the multirate system, checkerboard distortion may appear in a resultant image, thereby indicating that the filter designs that can be used are limited. That is, limitations are placed on filter designs in accordance with the images that are to be provided, and required filter conditions, such as the one according to which the filter property for a passing band and a blocking band must be flat, or a condition according to which a down sampling filter is separately provided to introduce an average operation. It is preferable that an FIR (Finite Impulse Response) filter be employed to maintain the linear phase of an image; however, usually, a higher order filter is required to obtain an image quality that is higher than that acquired by the bi-linear or bi-cubic interpolation method.
Conventionally, many improved methods have been proposed to correct outlines produced by bi-linear or bi-cubic interpolation that are blurred or stair-step shaped. However, for these methods it is premised that a boundary has at the least a specified size, and thus a font that has a structure consisting of one pixel width can not be satisfactorily coped with.
In reference Proceedings of the 1999 IEICE General Conference D-11-120, a resolution conversion method employing the re-formation of an oblique outline is proposed whereby the edge of an original image is detected, and an edge having an inclination that corresponds to that of the detected edge is re-formed for the resultant image, so that jaggies that are generated due to the expansion of an oblique line can be replaced on the resultant image with fine steps. According to this method, however, a 3×3 sized filter, at the least, must be employed to detect the edge, and the fine structure of a font can not be expanded.
In reference IEEE Proceedings of International Conference on Image Processing 1997, vol. I, pp. 267-270, October 1997, a method is described whereby linear expansion is employed for an edge that is enhanced in an original image. However, this method is not effective when a structure is smaller than the size of an edge detection filter, and outstanding stair-stepping jaggies are produced unless shaping is performed for an inclined boundary.
Further, in reference IEEE Transactions on Image Processing, vol. 8, no. 5, pp. 629-639, May 1999, a method is proposed for changing a distance weighing that is used for the interpolation of an outline. According to this method, a steep change on the original image is detected by four pixel points that are on both sides of a target pixel, and the distance from the interpolation point that is used for bi-linear or bi-cubic interpolation is changed. While this method is effective for obtaining a clear outline for a natural image, however, such as for a PC screen image, substantially it is only as effective as the nearest neighbor method.
SUMMARY OF THE INVENTIO
Sekiya Kazuo
Shimizu Junya
Cutter, Esq. Lawrence D.
Heslin Rothenberg Farley & & Mesiti P.C.
International Business Machines - Corporation
Johnson Timothy M.
Kassa Yosef
LandOfFree
Image conversion method, image processing apparatus, and... 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 conversion method, image processing apparatus, and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image conversion method, image processing apparatus, and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3192709