Facsimile and static presentation processing – Natural color facsimile – Specific image-processing circuitry
Reexamination Certificate
1998-08-24
2002-10-29
Grant, II, Jerome (Department: 2624)
Facsimile and static presentation processing
Natural color facsimile
Specific image-processing circuitry
C382S250000
Reexamination Certificate
active
06473207
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an image size transformation method for an orthogonal transformation coded image, particularly to that which enlarges and reduces the image size of coded data in respect with compression coding of an image using an orthogonal transformation.
DESCRIPTION OF THE RELATED ART
A conventional image size transformation method for an orthogonal transformation coded image is defined as a method for acquiring coded data of an enlarged or a reduced image size from coded data of an image compressed by an orthogonal transformation.
As to the image size transformation method for obtaining the coded data, the following image size transformation method is the most typical. The coded data, i.e. orthogonal transformation coefficients, is once reproduced into an image by an inversion. Then the image size is enlarged or reduced by interpolating or thinning out pixels at the image space, after which the image goes through compression coding by the application of the orthogonal transformation for the second time.
FIG. 1
shows a processing procedure of such enlargement process. Note that a reduction process would require the same processing procedure. In this conventional image size transformation method, however, not only enlargement/reduction process of the image size is required but also an inversion of the orthogonal transformation coefficients and a repetition of the. orthogonal transformation process are required. Accordingly, it is noted as a disadvantage that the whole image size transformation process would become complicated.
This sort of image size enlargement method is applied to a spatial scalable coding system in connection with the IS 13818-2 standard which is being internationally standardized by ISO/IEC JTCI/SC29/WG11. In this system, an image is coded with a low resolution, i.e. a small image size. The coding result is then referred to so as to code the image by a high resolution, i.e. a large image size. In coding a high resolution image, the data of the orthogonal transformation coefficients of the previously coded low resolution image is once inverted and brought back to a decoded image. This decoded image is then enlarged on the image space to become estimate data. Then with respect to each pixel of the high resolution image, difference data is obtained by making reference to the estimate data, after which the difference data is coded by the orthogonal transformation. In this system, the orthogonal transformation coefficients which are obtained by coding the low resolution image have to be once inverted, which means that an extra operation is required. Moreover, because an interpolation process using a band pass filter is required in enlarging the decoded image and generating the estimate data, there is a possibility that estimation errors should occur due to problems relating to operational precision and filtering characteristics of the interpolation process. In this respect, it becomes an issue that the coding efficiency is easily deteriorated.
Furthermore, as to another image size transformation method for enlarging or reducing a size of an image, there is known a method in which coefficients are added to or eliminated from in enlarging or reducing the image size. For example, according to an image enlargement method as disclosed in Japanese Patent Laid-Open Publication No. 2-76472, orthogonal transformation coefficients which are obtained by two dimentional discrete cosine transformation by a (N×N) point block unit are applied to a low frequency side of a (M×M) point block where M>N, while zero values are applied to a high frequency side of the block so as to obtain orthogonal transformation coefficients of (M×M) points. Then by inverting the orthogonal transformation coefficients of (M×M) points after transformation, by the two dimensional discrete cosine transformation of (M×M) points, it is possible to obtain a decoded image being enlarged by M/N times. On the other hand, by eliminating the high frequency side of the orthogonal transformation coefficients of (N×N) points, generating coefficients of a (K×K) point block where K<N, and inverting the coefficients by the two-dimentional discrete cosine transformation of (K×K) points, it is possible to obtain a decoded image being reduced by K/N times.
FIG. 2
shows a concept of such enlarging and reducing process. As to a similar image size transformation method, there is known a high quality image enlargement method disclosed in Japanese Patent Laid-Open Publication No. 3-204268. In the disclosure, instead of carrying out the process of applying zero values to the high frequency side as in the former method, values being anticipated by an autoregressive model are applied from the coefficients of the low frequency side. By the application of such process, it is possible to obtain a seemingly enlarged decoded image with high resolution components being reinforced.
However, in respect with the image size transformation method of the two prior art disclosures mentioned above, various magnifications and reduction ratios such as (M/N) time enlargement and (K/N) time reduction are possible. On the other hand, in decoding the orthogonal transformation coefficients after having the image size transformed, inverse orthogonal transformation of different sample points must be carried out depending on the magnification or reduction ratio. Accordingly, in respect with a decoding means which is usually provided beforehand with only a predetermined inverse orthogonal transformation function, there is a possibility that the coded data cannot be decoded. Furthermore, in case when a decoding device applicable for various magnifications and reduction ratios is made possible by a hardware circuit, it might be a problem that the circuit structure would become complicated.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a more efficient image size transformation method for an orthogonal transformation coded image, which is capable of enlarging and reducing a size of an image on orthogonal transformation coefficients, thus requiring a small operation amount.
According to a first aspect of the present invention there is provided an image size transformation method for an orthogonal transformation coded image, enlarging or reducing an image size of coded data, in respect with an image compressing and coding using an orthogonal transformation, comprising the steps of: reading out orthogonal transformation coefficients being the coded data; conducting a coefficient transformation by an operation process on the orthogonal transformation coefficients; and outputting the orthogonal transformation coefficients after having the image size enlarged or reduced, thus executing an enlargement or a reduction of the image.
According to a second aspect of the present invention, there is provided an image size transformation method for an orthogonal transformation coded image, enlarging an image size to output a decoded image, in respect with an image compressing and coding using an orthogonal transformation, comprising the steps of: reading out orthogonal transformation coefficients being coded by an N point orthogonal transformation, where N is a natural number; conducting a coefficient transformation by an operation process on the orthogonal transformation coefficients; calculating to obtain orthogonal transformation coefficients, after having the image size enlarged by a certain integer; and decoding the calculated transformation coefficients by an N point inverse orthogonal transformation so as to output an enlarged image, thus executing enlargement and decoding of a coded image.
According to a third aspect of the present invention, there is provided an image size transformation method for an orthogonal transformation coded image, coding an image into hierarchies to obtain a plurality of decoded images with different sizes, comprising a coding stage and a decoding stage. The coding stage includes the steps
Grant II Jerome
Whitham Curtis & Christofferson, P.C.
Worku Negussie
LandOfFree
Image size transformation method for orthogonal... 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 size transformation method for orthogonal..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image size transformation method for orthogonal... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2983196