Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
2000-07-17
2004-04-06
Chang, Jon (Department: 2623)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S224000, C382S261000
Reexamination Certificate
active
06718073
ABSTRACT:
TECHNICAL FIELD
The present invention relates to an arithmetic unit and conversion unit and methods thereof, and is suitably applied to a conversion unit for converting first data into second data with higher quality than the first data and to an arithmetic unit for computing prediction coefficients which are used for the conversion processing.
BACKGROUND ART
Under the situation where various kinds of digital units are available, a signal conversion unit that performs signal conversion is necessary for connecting units having different signal formats from each other. For example, in the case of displaying image data with low resolution on a monitor having high resolution, an image data conversion unit is necessary for producing image data with high resolution from image data with low resolution through format conversion. So far, an image data conversion unit of this kind has produced image data with high resolution through pixel interpolation in which frequency interpolation processing is performed on image data with low resolution by using an interpolation filter.
As an image data conversion unit, an up-converter adopting classification adaptive processing is used, in which image data with low resolution is classified into classes in accordance with a signal level distribution of pixels, and then prediction coefficients corresponding to the respective classes are read out from a memory in which the prediction coefficients are stored in order to obtain image data with high resolution from the prediction coefficients and the image data with low resolution by a predictive operation.
The prediction coefficients, which are stored in the memory, are generated in advance by data processing called learning. A learning circuit for generating the prediction coefficients, down-converts image data with high resolution as a teacher image with a digital filter to generate image data with low resolution as a pupil image and further generates the prediction coefficients by performing learning using the image data with high resolution and the image data with low resolution.
By the way, in the case where image data with high resolution has a plurality of signal characteristics, it is desirable that a frequency characteristic of the digital filter is changed according to each signal characteristic. That is, when image data with high resolution is generated from image data with low resolution, a digital filter whose frequency characteristic improves resolution is desirable for a still image portion since a resolving power of a human eye is improved for the still image portion, whereas a digital filter whose frequency characteristic suppresses improvement of resolution is desirable for a moving image portion since an unnecessary signal component in a high band is moved to deteriorate image quality as image data resolution is improved.
Accordingly, in the case where image data with high resolution is down-converted to image data with low resolution using one digital filter to generate prediction coefficients, prediction coefficients corresponding to the respective signal characteristics cannot be generated if the image data with high resolution has a plurality of signal characteristics. As a result, generating image data with high resolution from image data with low resolution cause a problem in which improvement of image quality is hindered.
Further, some cases may be better to change the frequency characteristics of a digital filter because of not only still image portions and moving image portions but also other factors.
DISCLOSE OF INVENTION
The present invention has been made in view of the above problem and it is an object of the present invention to provide an arithmetic unit and method for computing prediction coefficients which are used for conversion processing further matching the characteristics of first data as compared with conventional units and a conversion unit and method for generating second data using the prediction coefficients obtained by the arithmetic unit.
In order to solve the above problems, the present invention provides an arithmetic unit for computing prediction coefficients which are used to convert first data into second data having higher quality than the first data. The arithmetic unit comprises a class determining section for classifying teacher image data having higher quality than the first image data into a plurality of classes based on its characteristics, a pupil image data generating section for generating pupil image data having the same quality as the first image data by performing filtering processing different in each of classes determined by the class determining section, on the teacher image data, and a prediction coefficient generating section for generating prediction coefficients based on the pupil image data and the teacher image data.
The pupil image data matching the characteristics of the first image data and the teacher image is generated by performing filtering processing different in each class on the teacher image data. And the prediction coefficients matching the characteristics of the first image data and the teacher image data are generated by generating the prediction coefficients based on the pupil image data and the teacher image data.
Since the conversion unit uses the prediction coefficients based on the characteristics of the first image data, it can perform the conversion processing matching the characteristics of the first image data when converting the first image data into the second image data.
REFERENCES:
patent: 5379072 (1995-01-01), Kondo et al.
patent: 5488618 (1996-01-01), Kondo et al.
patent: 5666164 (1997-09-01), Kondo et al.
patent: 5748235 (1998-05-01), Kondo et al.
patent: 5852470 (1998-12-01), Kondo et al.
patent: 5946044 (1999-08-01), Kondo et al.
patent: 5966183 (1999-10-01), Kondo et al.
patent: 6181382 (2001-01-01), Kieu et al.
patent: 6285714 (2001-09-01), Kawaguchi et al.
patent: 0 912 045 (1999-04-01), None
patent: 0 708 5267 (1995-03-01), None
patent: 8-46934 (1996-02-01), None
patent: 08307835 (1996-11-01), None
patent: 8-317347 (1996-11-01), None
patent: 1 001 1583 (1998-01-01), None
patent: 10-112843 (1998-04-01), None
patent: 10-112844 (1998-04-01), None
patent: 10-191318 (1998-07-01), None
patent: 1 031 3251 (1998-11-01), None
Pohjala et al. “Line Rate Upconversion in IDTV Applications.” IEEE Trans. on Consumer Electronics, vol. 37, No. 3, Aug. 1991, pp. 309-312.
Horishi Takashi
Kondo Tetsujiro
Chang Jon
Frommer William S.
Frommer & Lawrence & Haug LLP
Sony Corporation
LandOfFree
Arithmetic device, and converter, and their methods does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Arithmetic device, and converter, and their methods, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Arithmetic device, and converter, and their methods will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3204915