Image analysis – Applications – Document or print quality inspection
Reexamination Certificate
1998-11-23
2002-02-26
Bella, Matthew C. (Department: 2621)
Image analysis
Applications
Document or print quality inspection
C382S254000
Reexamination Certificate
active
06351546
ABSTRACT:
This application is related to an application filed the same date and titled “Image Processing Method and Image Processing Device,” attorney docket no. JAO 36154A.
1. Field of the Invention
The present invention relates to an image processing method and an image processing device which evaluates objectively the copy quality of a copy machine, for example, enables restoration of a deteriorated image, and further may be used as a preparatory process for an optical character recognition (OCR) process.
2. Description of Related Art
The use of copy machines has spread widely in recent times. The capability of recent copy machines to restore images faithfully is improving rapidly. Digital copy machines, in particular, which are becoming more common today, can produce high quality copies with highly faithful copying capabilities.
Therefore, if an original to be copied has a clear and clean image, digital copy machines, with their highly faithful copying capabilities, can produce high quality copies that are virtually the same as the original. However, when the original is one that was copied by a conventional analog copy machine and the image has deteriorated, a digital copy machine, with its highly faithful copying capabilities, will reproduce a deteriorated copy.
Moreover, in recent years, optical character recognition technology is spreading in which a printed text is read by a scanner and characters from the input image data are extracted and changed into computer code. In such optical character recognition technology, a high ratio of recognition is achieved when a text is read by a scanner if the text is an original having a clear and clean character image as described in FIG.
41
(
a
) rather than a copy produced by a copy machine.
However, even if the original has a clear and clean image, the image deteriorates with repeated copying and, if such a deteriorated image is read by a scanner, the copy may result in deteriorated images as described in FIG.
41
(
b
) and FIG.
41
(
c
). FIG.
41
(
b
) represents an image produced by reading a once-copied original by a scanner, while (c) represents an image produced by reading the copy of a once-copied original (twice-copied original) by the scanner.
In order to cope with the deterioration of image quality as a result of copying, various measures have been provided for an optical character recognition mechanism. A dictionary for character recognition, including deteriorated character images, may be prepared and/or a function to repair some degree of unevenness provided.
As described above, a high level of capability to reproduce a faithful copy is required of copy machines, but when a deteriorated image (FIG.
41
(
b
) and FIG.
41
(
c
)) as a result of repeated copying is copied by a digital copy machine with a high level of capability to produce a faithful copy, there have been problems of inability to produce a high quality copy.
Moreover, evaluation of copy machine capability (evaluation of image quality) has been primarily performed by subjective evaluation. In other words, in general, human eyes, after seeing the copied image, determine the capability of the copy machine. Traditionally, there have been technologies associated with copy machines to correct the unevenness of deteriorated characters to some degree when characters which have deteriorated as a result of repeated copying are used as the object of optical character recognition or further copying, but there has been no technology to enable objective, numerical evaluation of image quality, to determine the quality of image, especially the so-called faintness or smudge of the characters, and to repair appropriately according to the deterioration condition of the image.
Moreover, in optical character recognition technology to extract and change into codes the characters from input image data, deteriorated characters are not sufficiently recognized. For example, suppose the character images of FIGS.
41
(
a
),
41
(
b
), and
41
(
c
) were targeted for character recognition. Assuming the character recognition rate of FIG.
41
(
a
) to be 100%, the rate drops rapidly with 90% for
41
(
b
) and 80% for
41
(
c
). Thus, a sufficient recognition rate is not obtained for deteriorated images. Moreover, images pleasing to the eye are not produced.
Furthermore, in assigning binary values to character images, it is necessary to determine the optimum binary threshold values, but the traditional binary threshold determination method does not result in optimum binary threshold values for characters. There have been cases in which faintness and smudges resulted after binary values were assigned to character images by final binary threshold values.
SUMMARY OF THE INVENTION
In order to solve these and other problems, the present invention provides an image processing method and an image processing device which, applied to a copy machine, evaluates the quality of the copied image objectively, determines deteriorated sections of the image, enables execution of an improvement process for the deteriorated sections according to the deterioration condition, enables determination of optimum binary threshold values for characters, and may be used effectively as a preparatory process for an optical character recognition device and the like.
An image processing method of the present invention comprises an image quality computation process wherein a characteristic amount is extracted to determine the image quality of image data entered by an image input device, and wherein the characteristic amount is computed as an evaluation value, and the image quality is determined by the evaluation value obtained by the image quality computation process.
Moreover, said computation process may include a first characteristic amount extraction process which has, to begin with, several patterns of pixel characteristics as characteristic points; computes the first characteristic amount, which is the ratio of the frequency of the appearance of said characteristic points in processing lines and the frequency of reversal of black pixels and white pixels; and determines image quality using the first evaluation value which is the first characteristic amount computed above.
Furthermore, said image quality computation process may include a second characteristic amount extraction process which computes the average length of a continuous string of black pixels nearly equivalent to the size of a character; computes the number of continuous strings of black pixels longer than the average continuous string of black pixels nearly equivalent to the size of a character; computes a second characteristic amount which is the ratio of the number of continuous strings of black pixels longer than the average length and one-half of the number of reversals of black pixel and white pixels; and determines image quality using a second evaluation value which is the second characteristic amount computed above.
In addition, said image quality computation process may include the first characteristic amount extraction process and said second characteristic amount extraction process; compute evaluation values based on a first evaluation value obtained by the first characteristic amount extraction process and a second evaluation value obtained by the second characteristic amount extraction process; and the image quality may be determined using an evaluation value based on the first and the second evaluation values.
Moreover, said image quality computation process may include a third characteristic amount extraction process to extract as the third characteristic amount the average length of a continuous string of black pixels nearly equivalent to a character in the processing line and in said first characteristic amount extraction amount; obtain an evaluation value based on a first evaluation value obtained from the first characteristic amount extraction process and a third characteristic amount obtained from the third characteristic amount extraction process; and the image quality may be determined using an evaluation value based on the first eval
Kobayashi Ayako
Kuwata Naoki
Murayama Yasuhiko
Oliff & Berridg,e PLC
Seiko Epson Corporation
LandOfFree
Image processing method and 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 method and image processing device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing method and image processing device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2968270