Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1998-06-12
2001-02-27
Lee, Thomas D. (Department: 2724)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S293000, C382S248000, C358S438000
Reexamination Certificate
active
06195472
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of Invention
This invention relates to image processing. In particular, this invention relates to image processing apparatus and methods that can rotate an image at any angle when orthogonal transform coding or decoding is performed.
2. Description of Related Art
Generally, an image contains an extremely large amount of data. When the image is stored or transmitted, it is often compressed by coding. Image coding methods are classified into two classes, i.e. lossless coding and lossy coding. In particular, the lossy coding often uses a technique in which pixel values are transformed into transform coefficients in a frequency domain. This is called a transform process. A decoded image is reconstructed from the frequency domain transform coefficients. This is called an inverse-transform process. One such lossy coding method is a discrete cosine transform, which is used in the well known JPEG (Joint Photographic Experts Group) standards.
There will now be explained a coding or decoding process using a frequency domain transform performed along with an image rotation process. The image rotation process is applied to pixel values in a spatial domain. A spatial domain can be, for example, the X-Y coordinate system. In a coding process, an image should be rotated before performing the frequency domain transform. In a decoding process, an image should be rotated after it is frequency domain inverse-transformed into a spatial domain. Thus, a memory is required to store temporarily the rotated image in the coding process or the inverse-transformed image before rotation in the decoding process. In addition, the total processing time of the coding or decoding process increases.
There is a process that performs a rotation process in a frequency domain without transforming an image into a spatial domain. A conventional rotation process in a frequency domain is disclosed by Japanese Laid-Open Patent Application No. 8-204957.
FIG. 15
is a block diagram illustrating the above-referenced conventional image processing apparatus performing a rotation process. The apparatus includes image input unit
42
, transform unit
52
, transform coefficient processor
90
, entropy encoder
61
and image output unit
71
. The reference numerals
141
,
151
,
159
and
160
represent an input image, transform coefficients, transform coefficients after image processing, and code data, respectively.
Image input unit
42
receives an input image from an outside source and outputs the input image
141
(as a plurality of blocks) to transform unit
52
. Transform unit
52
performs a frequency domain transform of each block in the input image, and outputs transform coefficients
151
to transform coefficient processor
90
. Transform coefficient processor
90
performs image processing of the received transform coefficients and outputs transform coefficients
159
after the image processing to entropy encoder
61
. Entropy encoder
61
performs entropy coding of the transform coefficients
159
after the image processing, and outputs code data
160
to image output unit
71
. Image output unit
71
outputs the code data of an image to a recording medium (memory), or the like. According to this configuration, the process of the conventional apparatus is explained with reference to
FIGS. 16A through 18
.
FIG. 16A
shows a conventional process that rotates an input image by an angle of 90°.
FIG. 16B
is a flowchart illustrating the steps performed by the conventional process shown in FIG.
16
A. In this example, (a block of) transform coefficients
151
are rotated at an angle of 90° so as to obtain (a block of) transform coefficients
159
. In step S
200
of
FIG. 16B
, transform coefficients
152
in each block are calculated from transform coefficients
151
by replacing a transform coefficient of coordinate (u, v) in
151
with that of coordinate (v, u) (see
FIG. 17A
). This is referred to as rotation about the u-v axis. In step S
210
, transform coefficients
159
are calculated by inverting the signs of the transform coefficients
152
located at the odd coordinates u in the block of the transform bases (see
FIG. 17B
in which the darkened blocks identify the transform coefficient
152
at the odd coordinates). In step S
220
, each block of (X, Y) is moved to the position (N+1−X, Y) (see FIG.
17
C). (X, Y) indicates a position of a block in an image. N indicates the number of blocks in the X direction. Thus, if an image is segmented into a 12×12 matrix of blocks, there are 144 blocks, and N=12. Thus, for example, the block at position (1,1) will be shifted to position (12+1−1,1)=(12,1). As a result of this process, an image is rotated at an angle of 90°. In addition, as shown in
FIG. 18
, an image can be rotated at an angle of 180° or 270° by performing different combinations of processes similar to the processes described above. (With respect to correlating
FIG. 18
to
FIG. 15
, numerals
153
and
154
are transform coefficients like numeral
151
; and numerals
157
and
158
are transform coefficients like numeral
159
.) As shown in
FIG. 18
, to rotate by 180°, perform a top-bottom symmetrical reversing process followed by a left-right symmetrical reversing process; to rotate by 270°, perform a rotation followed by a top-bottom symmetrical reversing process.
As described above, the conventional process can rotate an image at an angle of 90°, 180° or 270° by performing image processing in the frequency domain. Thus, the rotation process can be performed in a frequency domain. However, this method cannot rotate an image at any angle other than 90°, 180° or 270°.
SUMMARY OF THE INVENTION
The present invention has been made in view of the above circumstances. One aspect of the present invention provides an image processing apparatus and method that satisfies the following goals. When a frequency transform is performed in image coding along with a rotation process, a rotated coded image can be obtained without creating a rotated image in a spatial domain. Meanwhile, when a frequency inverse-transform is performed in image decoding along with a rotation process, a rotated decoded image can be obtained without creating a decoded image prior to rotation. In addition, an image can be rotated at any arbitrary angle.
To achieve the above and/or other goals, the present invention provides an image processing apparatus and method that inputs a rotation angle. According to the input rotation angle, transform bases and coordinates in an input image are calculated. A part (e.g., a block) of the input image is input based on the calculated coordinates. Then, the part of the input image is transformed into transform coefficients by using the calculated transform bases. Entropy coding is performed for the transform coefficients. This process is performed for each part of the image.
According to the image processing apparatus and method of the present invention, since an input image is transformed into transform coefficients by using transform bases that are rotated at an arbitrary angle, the transform coefficients become the same as the transform coefficients that are obtained by transforming a rotated image. Accordingly, a coded image rotated at an arbitrary angle can be obtained without creating a rotated image in the spatial domain. As used herein, “arbitrary” angle means any angle, not only 90°, 180 and 270°.
The present invention further provides an image processing apparatus that inputs a rotation angle. According to the input rotation angle, transform bases and coordinates in an output image are calculated. Entropy decoding is performed for an input transform-coded image so as to obtain transform coefficients. The transform coefficients are inverse-transformed into pixel values by using the calculated transform bases. Then, the pixel values are output in accordance with the calculated coordinates in an output image.
According to the image processing apparatus and method of the present invention, since an inverse-transform is per
So Ikken
Yokose Taro
Fuji 'Xerox Co., Ltd.
Lee Thomas D.
Oliff & Berridg,e PLC
LandOfFree
Image processing apparatus and method 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 apparatus and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing apparatus and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2605833