Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
1998-04-06
2001-02-20
Lee, Thomas D. (Department: 2724)
Image analysis
Pattern recognition
Feature extraction
C382S195000
Reexamination Certificate
active
06192153
ABSTRACT:
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing device constituting an electric file, digital copying machine, facsimile device, scanner and the like, and more particularly, to an image processing device for performing the best processing to each image signal gained by scanning a document, according to the characteristics of each pixel.
BACKGROUND OF THE INVENTION
In digital copying machines and facsimile devices, an image processing device for performing image processing against image signals gained by reading a document comprising a letter portion, a photograph portion or a dot portion, or the mixture of these plurality of portions by CCD image sensors and the like to improve the image quality of the recording image. This image processing is performed by identifying whether the read image type is either a letter image, a photograph image, or a dot image, and according to the identification result, performing the best processing to each pixel.
For example, Japanese Laid-Open Patent Publication No. 6-150059 discloses an invention where document image is divided into blocks comprising a plurality of pixels, and the image type of the document image is identified by matching patterns. In this invention, a plurality of basic patterns are prepared, and by comparing the document image pattern with the basic patterns, the image type of the document image is decided.
Further, in Japanese Laid-Open Patent Publication No. 8-149297, an invention for identifying whether an observation pixel is included in a dot image portion or not by utilizing the periodicity of the image data of dot images and detecting the periodicity of the image data of a line in the document image is disclosed.
Still further, Japanese Laid-Open Patent Publication No. 61-194968 discloses a dot photograph region identification method wherein a pixel inside a block having different signal levels in the main scan direction and the sub scan direction is measured, and by comparing the sum of the change coefficient, identifying whether the observation pixel is of a dot photograph image portion or not.
However, in the image identifying process by matching patterns as disclosed in Japanese Laid-Open Patent Publication No. 6-150059, it is necessary to prepare many patterns beforehand, which leads to the need of a large memory capacity, and less universality.
Further, in the method of detecting the periodicity of the image data of one line as disclosed in Japanese Laid-Open Patent Publication No. 8-149297, there are cases where the dot portion could not be identified reliably, since in the case of a dot region with low density and small dot sizes, dots are rarely placed on one line. Therefore, it is necessary to detect periodicity of the image by plurality of lines in two dimension regions, which leads to necessity of many block memories according to the periodicity of the dots, which further leads to rise of costs.
Even further, according to the invention where continuity of dot photograph portions is detected as is disclosed in Japanese Laid-Open Patent Publication No. 61-194968, the detection of dot regions may be possible, but the identification between the photograph portion having small difference in signal levels with the letter portion is difficult, and accurate identification of dot photograph portions could not be realized.
Therefore, the object of the present invention is to provide an image processing device enabling a reliable identification of whether which pixel of the document image is included in which of the plurality of image regions by a small circuit structure.
SUMMARY OF THE INVENTION
(1) It is an object of the present invention to provide an image processing device for identifying an image region including each pixel of a document image, and performing a process for each pixel according to the identification result of the image region, wherein the image processing device comprises an edge detect means for detecting the pixels positioned in an edge portion of an image by calculating a sum of products of each image signal of a local block comprising an observation pixel and a plurality of peripheral pixels with said observation pixel positioned at the center thereof multiplied by a filter coefficient having a sum of −1, and an identification means for identifying the image region including said observation pixel based on the detection result of said edge detect means.
In said image processing device, the edge pixel positioned at the edge portion of an image is detected by calculating a sum of products of each image signal of a local block comprising an observation pixel and a plurality of peripheral pixels with said observation pixel at the center multiplied by a filter coefficient having a sum of −1. When a filter whose sum of filter coefficient is −1 is used for edge detection, only the edge portion of the low density regions such as the letter region and the dot region of the document is detected, and no edge portions would be detected at the photograph regions. Therefore, by performing the edge detection to a plurality of pixels included in the local block using a filter whose sum of filter coefficient is −1, it could be identified whether the observation pixel is of an image of a low density region.
(2) It is another object of the present invention to provide an image processing device characterized in that said identification means identifies whether the observation pixel is included in a photograph region based on the binary coded data of the detection results of the edge detect means by a predetermined threshold.
In said image processing device, the image signal intensified by using a filter whose sum of filter coefficient is −1 for a plurality of pixels included in the local block is binary coded by a predetermined threshold in order to identify accurately whether the observation pixel is included in the photograph region.
(3) It is another object of the present invention to provide an image processing device characterized in that said identification means calculates a continuation number of the pixels positioned at the edge portion in the local block, and based on this calculated result, identifies whether the observation pixel is included in a photograph region, a letter region, or a dot region.
In said image processing device, the continuation number of the edge portion pixels detected in the local block with the observation pixel of a low density region which is either a letter region or a dot region positioned at the center thereof is counted. At the local block with the observation pixel included in the photograph region, no edge pixel could be detected, and at the local block with the observation pixel included in the letter region, the continuation number of the edge portion pixels is large, and at the local block with the observation pixel included in the dot region, the continuation number of the edge portion pixels is small. Therefore, whether the observation pixel is included in a photograph region, a letter region, or a dot region could be identified accurately by the continuation number of the edge portion pixel in the local block including the observation pixel being either 0, large value, or small value.
(4) It is another object of the present invention to provide an image processing device characterized in that said identification means identifies whether or not the observation pixel is included in a picture region based on the binary coded data of the detection result of the edge detect means by a predetermined threshold, when the observation pixel is not included in the photograph region, calculates the continuation number of the pixels positioned at the edge portion in the local block and based on the calculation results, identifies whether the observation pixel is included in a letter region or a dot region.
In said image processing device, when no edge portion pixels exist in the local block of the observation pixel, and the observation pixel is included in the picture region, the continuation number of the
Hayashi Yoshinori
Morita Kiwamu
Suzuki Michiyuki
Brinich Stephen
Lee Thomas D.
Sharp Kabushiki Kaisha
LandOfFree
Image processing device 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 device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2581656