Image analysis – Image transformation or preprocessing – Measuring image properties
Reexamination Certificate
1997-09-30
2001-01-09
Lee, Cheukfan (Department: 2722)
Image analysis
Image transformation or preprocessing
Measuring image properties
C382S290000, C358S488000
Reexamination Certificate
active
06173088
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an image forming method and apparatus, which input an original image and form an image on a recording medium.
A conventional image forming apparatus such as a copying machine forms images on recording sheets in the directions the original images are read. For this reason, when a bundle of originals include both horizontally and vertically written documents or have different image directions, a bundle of output sheets also have different directions. This is beginning to pose a serious problem along with development of a post-process mechanism such as bookbinding.
To solve this problem, a system for recognizing original image directions using a character recognition technique, and adjusting the directions of output images to a given direction has been proposed. The recognition technique of the original image direction uses a technique prevalent in an information processing apparatus. However, the information processing apparatus selects character candidates based on features obtained assuming that a character points in a true direction. For this reason, when image data rotated 90°, 180°, or the like is read, character recognition cannot be normally done, and a quite different character code is often output as a recognition result. For example, in the case of a character
shown in
FIGS. 1A
to
1
D, when the character points in the true direction (FIG. 1A),
is normally recognized. However, when the character is rotated 270° clockwise (FIG. lB),
is erroneously recognized; when the character is rotated 180°(FIG.
1
C),
is erroneously recognized; and when the character is rotated 90° (FIG.
1
D), recognition is disabled (note that
FIGS. 1A
to
1
D show imaginary character recognition results for descriptive purposes only, which do not always become as indicated in practice).
In order to perform normal character recognition, when the document direction is not true, the direction of an original is manually corrected, and the original is read again to execute character recognition. However, since the processing speed of a scanner has improved, and an automatic document feed function called an auto feeder is often exploited, a large number of originals are likely to be processed at one time, and it is hard to correct original directions manually. On the other hand, there are several types of original document formats such as a vertically written document, horizontally written document, and the like: more specifically, an A4 horizontally written portrait document (see FIG.
2
A: popularly used for a Japanese horizontally written document, an English document, and the like), an A4 horizontally written landscape document (see FIG.
2
B: popularly used for a document with long sentences, an OHP document, a document obtained by copying A3 or B4 documents in a reduced scale, and the like), an A4 landscape document in which the column directions are switched at the center of the page (see FIG.
2
C: popularly used when two A4 documents are continuously copied in a reduced scale), an A4 vertically written portrait document (see FIG.
2
D), and the like.
On the other hand, the direction of an image displayed on a monitor is determined by the scanning direction (image reading direction) of a scanner upon inputting an image. For this reason, when an original is read in a direction in which the scanning direction of the scanner agrees with the direction of a character string on a document, the document is displayed in its true direction, as shown in FIG.
3
A. However, in the case of an A4 scanner, since the original direction is uniquely determined, the scanning direction of the scanner does not agree with the direction of a character string on a document, and the character string on the document is displayed while it is rotated 90° from its true direction, as shown in, e.g.,
FIG. 3B
, resulting in an image which is very hard to read.
In order to avoid the above-mentioned recognition errors of a character and an image which is hard to read, an information processing apparatus with an automatic document image direction determination/rotation function has been conventionally proposed. In typical ones of conventional automatic document image direction determination schemes, as shown in
FIGS. 4A and 4B
, the direction of a portion with lines such as a table is checked based on the region division result, and the direction that divides character regions with a horizontally elongated line (see “a” in
FIG. 4A
) is determined to be the direction of that document image. Also, histograms in the vertical and horizontal directions of a document are detected, and the direction of the document is determined by checking peaks in the detected histograms (see “b” in FIG.
4
B). Furthermore, character regions are extracted from a document, and the direction of the document is determined by checking the features of rectangular regions such as a horizontally or vertically elongated region that matches the feature of a character region (see “c” in FIG.
4
B). A document image is rotated based on the document image direction determined by such scheme, and character recognition is done for the rotated image. In particular, such correct character recognition based on the document direction has been increasingly anticipated along with applications of electronic filing for arranging documents in large quantity, development of DTP, and development of equipment such as copying machines that can process documents in large quantity, and techniques that can perform recognition more accurately have been proposed.
However, in the prior art, the information processing apparatus detects the directions of original images using character recognition, and automatically rotates images to adjust the image directions and to control the stapling directions. However, when originals include those having columns or figures, character recognition cannot be accurately done. As a result, original image directions are erroneously detected due to the presence of various types of originals, and copies that the operator did not intend are produced.
The copying machine improves its operation efficiency by attaching accessories such as an automatic document feeder, automatic sorter, and the like. A digital copying machine has been proposed, and sophisticated functions can be realized since an image is processed as digital information. Such functions include, e.g., a so-called reduced-scale layout function.
FIG. 5
shows an example as a combination of a two-sided copying function, binding margin function, and stapling function with this reduced-scale layout function. In
FIG. 5
, reference numerals
1201
,
1202
,
1203
,
1204
, . . . denote a plurality of horizontally written portrait originals. When these originals are set on an automatic document feeder (DF) in the illustrated direction, images on these originals are copied in a layout
1205
on the front side and in a layout
1206
on the back side to have a binding margin, and output copies are stapled at a desired position. Note that “1”, “2”, “3”, “4”, . . . , “8” in
FIG. 5
correspond to the order of originals, i.e., the pages of originals.
However, the above-mentioned prior art suffers the following problem. For example, when vertically written portrait originals
1207
,
1208
,
1209
,
1210
, . . . are set, images on these originals are copied in a layout
1211
on the front side and in a layout
1212
on the back side, resulting in a very unnatural layout for vertically written documents. Furthermore, the binding margin position and staple position are not suitable for a practical use.
Similarly, when vertically written landscape originals
1213
,
1214
,
1215
,
1216
, . . . are set, images on these originals are undesirably copied in a layout
1217
on the front side and in a layout
1218
on the back side. Also, when horizontally written landscape originals
1219
,
1220
,
1221
,
1222
, . . . are set, images on these originals are undesirably copied in a layout
1223
on the front side and in a layout
1224
Funada Masahiro
Koh Shokyo
Toda Yukari
Yagasaki Toshiaki
Canon Kabushiki Kaisha
Fitzpatrick ,Cella, Harper & Scinto
Lee Cheukfan
LandOfFree
Image forming method and apparatus 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 forming method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image forming method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2557623