Image analysis – Histogram processing
Reexamination Certificate
1999-07-09
2004-09-28
Grant, II, Jerome (Department: 2626)
Image analysis
Histogram processing
C382S169000, C382S170000, C382S171000, C358S483000
Reexamination Certificate
active
06798905
ABSTRACT:
This application is based on applications No. 10-195995 and No. 10-250124 filed in Japan, the contents of which are hereby incorporated by reference.
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The present invention relates to a document orientation recognizing device which is usually provided in an image forming apparatus such as a copier and which recognizes the orientation of an image of a document.
(2) Related Art
There have been research and development in techniques relating to copying multiple documents successively using copiers, especially digital copiers, in such a manner that recording sheets will have respective images thereon in the same orientation regardless of random orientations of documents originally having placed by a user. An example of such a technique is disclosed by Japanese Laid-Open Patent Application No. H06-103710.
When the user sets documents arranged in random orientations on the copier for successive copying operations, the orientations of images formed on the discharged recording sheets (the recording sheets on which images have been formed are referred to as the “copied sheets” hereinafter) are also random in the same way as the documents having set by the user. This means that the user has to arrange the documents in the same orientation before the copying operations or change the orientations of the copied sheets after the copying operations, and this causes great inconvenience.
To align the orientations of images formed on copied sheets, the orientation needs to be detected for each document in accordance with image data of the document (referred to as the “document image data”) and image rotation processing may be performed when necessary. Hereinafter, the processing for judging the orientation of a document image is referred to as the “orientation detection processing.” In most of orientation detection methods, the orientation of one character retrieved from document image data is detected and that orientation is judged as the orientation of a whole document image. The image rotation processing is performed when the orientation of the document image that has been detected through the orientation detection processing disagrees with a predetermined orientation. Specifically, the image data is rotated a necessary angle so that the orientation of the document image data agrees with the predetermined orientation. By executing image formation based on the rotated image data, the orientations of images formed on copied sheets are oriented in the same direction.
For the orientation detection processing, various methods have been suggested with the aim of increasing processing efficiency and reliability of detection results. As one example, Japanese Laid-Open Patent Application No. H09-69136 discloses a method for increasing the reliability of detection results.
The orientation detection processing is performed on a basic premise that the orientation of characters in a document is identical to the orientation of a whole document image. The aim of the method disclosed in the Application H09-69136 is to reduce detection errors which occur when the orientation detection processing is performed on exceptional characters that exist on documents.
FIGS. 1
a
and
1
b
are examples where the orientations of characters on documents do not correspond to the respective orientations of whole document images.
A character string
801
in
FIG. 1
a
is a caption. The “caption” is appended to a graph or figure and explains what the graph or figure is about. The caption characters of the character string
801
are written horizontally in a vertical line, although a document
810
is in portrait format.
A character string
802
in
FIG. 1
b
is composed of characters included in a table (referred to as the “table characters” hereinafter). The character string
802
is written horizontally in a vertical line, although a document
820
is in portrait format.
It can be readily understood that the orientation of the document image is incorrectly detected when the orientation detection processing is performed on the caption characters or table characters. To address this problem, the cited Application No. H09-69136 discloses a method by which the orientation detection processing based on the caption characters and table characters are avoided when possible. This method is achieved according to the following steps.
First, an image part of a document is divided into a plurality of areas. The attribute of each area is next judged. The attribute is judged to be one of “text attribute,” “title attribute,” “table attribute,” and “caption attribute.” The text attribute and title attribute respectively indicate that a character in the area belongs to a text and a title. The table character attribute indicates that a character in the area belongs to a table, while the caption attribute indicates a character in the area belongs to the caption explaining what the figure or the like included in the document is about. According to the attribute judged for each area, priorities are respectively assigned to the areas for detecting the orientation of the document. In general, the text and title attributes take precedence over the table and caption attributes. As one example, suppose that the order of priority is text attribute →title attribute→table attribute→caption attribute. A plurality of characters are sampled from the area with the attribute of the highest priority, and the orientation detection processing is performed for each of the sample characters. When the results obtained through the orientation detection processing are the same, that is, the orientations of the sample characters judged to be the same, the orientation of those characters is judged as the orientation of the whole document image. On the other hand, when the orientations of the sample characters are different to each other, another plurality of characters are sampled from the area with the attribute of the second highest priority, and the orientation detection processing is performed for each of the sample characters.
Using this conventional technique, however, the attribute is judged for each area and the orientation detection processing is performed for each area in order of priority. This causes a considerable load. The processing performed in this way improves the precision of an orientation detection result and so is surely useful. Yet, in reality, the order of priority is fixed and the orientation detection processing is performed based on the area with the text or title attribute in most cases. The orientation detection processing may be performed based on the area with the table or caption attribute when the document has these kinds of characters only. In this case, the reliability of the orientation detection result would be low using any orientation detection method. For a document which includes both text and caption, it is likely for the orientation detection processing to be performed based on the text, not on the caption with a low priority. If there should be a case where the orientation detection processing is performed based on the caption instead of the text, the case would be very rare. Accordingly, in most cases, the load caused by the orientation detection processing performed in order of priority that is determined for the areas according to their attributes is too heavy for the effect obtained through this processing.
Meanwhile, there have been copiers which are provided with a mode where a double-page spread from a bound document such as a book is copied page by page. Hereinafter, this mode is referred to as the “page-by-page mode,” and bound documents, such as books and magazines, are all referred to as “books.” This mode is used when the user makes copies of a book, turning pages one by one. Here, it should be noted that the Japanese language can be written both horizontally and vertically. When the Japanese script is written vertically, it is written from right to left. This means that a book written vertically in Japanese is read in the opposite direction
Imaizumi Shoji
Sugiura Hiroshi
Ueda Kazuhiro
Grant II Jerome
McDermott Will & Emery LLP
Minolta Co. , Ltd.
LandOfFree
Document orientation recognizing device which recognizes... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Document orientation recognizing device which recognizes..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Document orientation recognizing device which recognizes... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3261497