Image analysis – Image compression or coding – Gray level to binary coding
Patent
1997-09-17
2000-01-11
Rogers, Scott
Image analysis
Image compression or coding
Gray level to binary coding
382272, 382273, 358429, 358455, 358466, G06K 936, H04N 140
Patent
active
060144653
DESCRIPTION:
BRIEF SUMMARY
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method for transforming a gray-level image into a black-and-white image, and in particular for transforming an image that has been read by a scanner.
PRIOR ART
A gray-level image is normally considered as a number of pixels which all together cover the complete image area. To each picture element, also called pixel, there is associated an intensity value. This value is a digital value, often several bits wide. If, by way of example, the intensity value is 8 bits wide then it can be represented by one of 2.sup.8 =256 possible different values, or so called gray-levels.
Image data in gray-level form can be obtained from a number of different equipments such as video equipments, conventional scanner machines, laser scanners, fax machines or computers.
In particular, when reading images or documents by means of a scanner or image reader, signals with digital values of 8 bits are normally produced. A scanner illuminates the image or document and examines the intensity of the reflected light in a large number of points that are arranged regularly in a point raster over the surface. Each such discrete point is represented by a pixel on the gray-level image, and a quantized intensity value, a gray-level, is produced depending upon the measured reflectivity of the point.
The image processing executed on these image data often starts with a transformation into an image representation of only two different intensity values, black (0) or white (1), i.e. a binary image. This gives an often well needed reduction of data (by a factor of 8) and efficient transmission and/or compact storing of image data can be obtained by further compression using special algorithms for images with binary intensity values. There even exist specific compression algorithms for this type of images. CCITT has standardized this type of algorithms for use in telefax technology.
In most cases, the transformation is accomplished by comparing intensity values of the original image with a threshold value, intensities over the threshold level being classified as white and intensities below the same level as black.
However, the reflected light varies in strength due to a number of reasons, and in practice there exist not only two light intensities on a typical image- or document surface. Even though a document is evenly illuminated and consists of clean white paper with black printing, the light measurement will sometimes, e.g. when a contour falls within the aperture, give rise to levels between black and white. The scanner does not measure the light in an infinitely small point, but always examines a surface area of a certain extent. If the threshold value lies closer to black than white, then, by way of example, a document that has been read by a scanner and transformed into binary form may exhibit text that is "thinner" than on the original, with a risk for interruptions in thin sections. Correspondingly, the text may become "fatter" with a risk for bridging between characters if the threshold value lies too far in the other direction. Accordingly, an incorrectly selected threshold value may deteriorate the readability of a document and seriously aggravate OCR (Optical Character Recognition) . Consequently, the method used for transforming gray-level images into black-and-white images is of considerable importance for the quality and readability of a transformed document.
Known methods for finding an appropriate threshold value are often based on examining of the complete image area. A simple principle is to set the threshold value midway between the greatest value and the smallest value that are found. A more sophisticated method is to construct, in principle, a histogram with a bar for each one of a number of partial intervals, where the height of the bar indicates the number of pixels that have values within the interval. For a typical black-and-white document this histogram should have two peaks, whereby the threshold value is selected midway between the peaks. These meth
REFERENCES:
patent: 4837846 (1989-06-01), Oyabu et al.
patent: 4850029 (1989-07-01), Moyer et al.
patent: 4903316 (1990-02-01), Hongo et al.
patent: 5608814 (1997-03-01), Gilmore et al.
patent: 5838463 (1998-11-01), Gahang
Rafael C. Gonzales et al., "Digital Image Processing", 1993, Addison-Wesley Publishing Company, p. 183, line 7-line 22; p. 443-p. 444.
Patent Abstracts of Japan, vol. 18, No. 69, E-1502, abstract of JP, A, 5-284356 (Seiko Epson Corp), Oct. 29, 1993.
Patent Abstracts of Japan, vol. 5, No. 55, E-52, abstract of JP, A, 56-6523 (Hitachi Seisakusho K.K.), Jan. 23, 1981.
Blixt Christian Bjorn Stefan
Blixt Stefan Sven
Rogers Scott
Stefan Christian Bjorn
LandOfFree
Method for transforming a gray-level image into a black-and-whit does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for transforming a gray-level image into a black-and-whit, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for transforming a gray-level image into a black-and-whit will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1468211