Image analysis – Image transformation or preprocessing – Combining image portions
Reexamination Certificate
1999-07-02
2002-04-30
Mehta, Bhavesh (Department: 2621)
Image analysis
Image transformation or preprocessing
Combining image portions
C382S294000, C382S224000, C358S450000
Reexamination Certificate
active
06381376
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an image processing device which outputs a plurality of input images as a single image by connecting the input images in such a manner as to overlap partially.
BACKGROUND OF THE INVENTION
One type of conventional image forming apparatuses, such as a copying machine, includes an image processing device which forms a single image by connecting a plurality of input images in such a manner as to overlap partially, so that the image forming apparatus of this type forms, in other words, prints out on a recording medium (e.g., a paper sheet) an image based on the image formed by the image processing device. When a copying machine makes, on a recording medium, a copy of an original larger than a scanner size which is defined as the largest readable size of an image reading section, the image processing device equipped therein takes the following procedure, for example.
In the first place, an original image is divided into a plurality of sub-images in such a manner as to partially overlap and the sub-images are read separately. Then, the sub-images thus read are connected to restore the original image, and the size of the restored original image is reduced to the size of a recording medium, after which the resorted original image is outputted onto the recording medium.
For example, Japanese Laid-open Patent Application No. 314263/1992 (Japanese Official Gazette, Tokukaihei No. 4-314263, published on Nov. 5, 1992) discloses an image processing device which functions in the following manner. That is, binary conversion processing and edge extraction processing with the binary image are conducted for each of a plurality of input images. Subsequently, the input images are compared one with another by means of pattern matching. Then, the input images are connected in such a manner that their edges having matching characteristics are overlapped.
According to the arrangement disclosed in the above publication, characteristics data of the input images obtained by the binary conversion processing followed by the edge extraction processing are compared by means of pattern matching. Thus, in case of text images composed of characters, image characteristics can be readily obtained by the edge extraction processing following the binary conversion processing, and therefore, two input images can be connected to each other in a relatively accurate manner. However, in case of shadow images or picture images, image characteristics can not be easily obtained by the edge extraction processing following the binary conversion processing, and therefore, matching portions of the two input images can not be readily found from comparison by means of pattern matching. Hence, two input image can not be connected accurately, and there arises a problem that a plurality of input images can not be resorted to a single image.
SUMMARY OF THE INVENTION
The present invention is devised to solve the above problems, and therefore, has an object to provide an image processing device which can restore a single image by connecting a plurality of input images accurately whether the input images are character, shadow or picture images.
In order to fulfill the above and other objects, an image processing device of the present invention is characterized by comprising:
a matching data generating section for generating matching data for each pixel in each of a plurality of input images by using multi-value image data thereof, the matching data representing characteristics of each pixel;
a region classifying section for classifying each pixel into one of a plurality of regions, which have been defined in advance based on kinds of images, by using the multi-value image data, the area classifying section also appending identification data exhibiting a result of classification to data of each pixel;
a binary converting section for converting the multi-value image data processed by the region classifying section into binary image data;
a pattern matching processing section for extracting a matching portion from data of each input image processed by the binary conversion section based on the identification data and matching data; and
a connection processing section for connecting the binary image data of each input image in such a manner that the matching portions extracted by the pattern matching processing section overlap.
According to the above arrangement, the matching data generating section generates the matching data for each pixel in each of a plurality of input images by using multi-value image data thereof so as to represent characteristics of each pixel. The region classifying section classifies each pixel into one of a plurality of regions, which have been defined in advance based on kinds of images, by using the multi-value image data, and also appends identification data exhibiting a result of classification to data of each pixel. The binary converting section converts the multi-value image data processed by the region classifying section into binary image data. The pattern matching processing section extracts a matching portion from data of each input image processed by the binary conversion section based on the identification data and matching data. The connection processing section connects the binary image data of each input image in such a manner that the matching portions extracted by the pattern matching processing section overlap.
Thus, according to the image processing device of the present invention, the pattern matching processing is conducted adequately depending on the kind of region to which each pixel belongs, and the matching portions of the input images can be extracted based on such pattern matching processing by using the matching data of each pixel generated in detail from the multi-value image data and the identification data of the region into which the pixel is classified in a reliable manner based on the multi-value image data. Consequently, a plurality of input images can be connected accurately to restore a single image whether the input images belong to any of the pre-defined regions including the character, shadow, picture, and background regions, for example.
For a fuller understanding of the nature and advantages of the invention, reference should be made to the ensuing detailed description taken in conjunction with the accompanying drawings.
REFERENCES:
patent: 5465163 (1995-11-01), Yoshihara et al.
patent: 6002492 (1999-12-01), Kamon et al.
patent: 6148118 (2000-11-01), Murakami et al.
patent: 6259826 (2001-07-01), Pollard et al.
Conlin David G.
Dike Bronstein Roberts & Cushman
Mehta Bhavesh
Sharp Kabushiki Kaisha
LandOfFree
Restoring a single image by connecting a plurality of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Restoring a single image by connecting a plurality of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Restoring a single image by connecting a plurality of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2931081