Method for facsimile recognition and reproduction of printed product text

FIELD: polygraphy.

SUBSTANCE: method includes conversion of recognized and standard images to digital form, their digital processing by determining coordinates, comparison and determining of match of recognized and standard contours. Determining of coordinates of line of characteristic contour of recognized image of symbol is performed using appropriate standard graphic image by finding value of coordinates X, Y, angle β of position of optical center of text symbols by superposition along area of printed area of digital images - in straight contrast of standard on appropriate recognized in reversed contrast.

EFFECT: higher reliability.

2 cl, 1 dwg

The invention relates to the field of printing, in particular to a computer system of text processing and text finals by means of interactive graphics in facsimile reproduction (reproduction) text sample job printed products.

In General printing, depending on the purpose and use of samples of printed products, as well as the requirements for the degree of facsimiles (correspondence) received printed edition prints of the originals (the standards) are applied to various systems and methods for automatic and semi-automatic recognition and playback of graphemes characters with the corresponding create text finals and receive printed sheets from one printing methods.

According to theoretical fundamentals of pattern recognition (Gorelik A.P., Skrypnik V.A. Methods of recognition. M., High school, 1977) and the requirements for fidelity geometric parameters of the image on the printing reproduction (Rahimova E.A. Control the size and shape of the images in photoreproduction processes. R.ZH Printing industry. Issue 1, p.40, Moscow, "Book", 1975) criteria facsimiles are distinctive thresholds. Such thresholds are calculated from the conditions of comparative evaluation in visual perception, reproduction and the origin is a, in izmenami thicknesses of the strokes are ±5-10 microns.

In the known facsimile method facsimile automated processing and reproduction of newspaper text are in the technological cycle "print with an inlaid form (original) → prints (film Assembly) → edition prints with matrices (stereotypes)" (Witt A.A. Determination of optimum process parameters decentralized production of Newspapers using facsimile transmission. Abstract of PhD thesis. M, 1971). The method consists in the fact that the copy, printed from the original form of the text set and rasterized images, scanned in analog mode in the longitudinal or personnel sweep direction. The signals from the photoelectric Converter after the formation and transformation on the spectrum are served in a communication channel with a frequency of 240 kHz. In the reception of electrical signals control the action of a discharge lamp, exposing the photographic material. Received after photographic processing facsimile negative photocopy is used for the manufacture of printed forms or her stereotypes and receive the printed sheets. For assessment criterion fidelity font on their printed point adopted the absolute value of the deformation of the vertical and horizontal resolution is Talnah strokes. Such deviations even when playing high-quality reprint of the original coated paper are ±13% (20-40 microns) and are beyond the thresholds fuzzy, which breaks in a separate headsets font unity build. The reason for the absolute deformation of the image elements of the characters of the font are systematic aperture distortion that occurs when an electron-optical analysis and image recognition deploy element, as well as the irregularity of amplitude-frequency characteristics, noise, radio interference, and random variance of the instantaneous velocities of the transmitter and receiver. This method ensures that facsimiles on the drawing sheets with matrices (stereotypes) only General typeface and within thresholds and zones fuzzy parameters such as line spacing and justification of rows, the height and width of the bands of the text set.

The currently existing methods of automatic image recognition based on the use of the relevant reference images (Patent (19) EPO (EP), (11) application No. 0114248, CL MKI G 06 To 9/78, UDC 681.32, publ. 84.08.01, No. 31, patent (19) EPO (EP), (11) application No. 115462, (51) MCI G 06 To 9/30, (53) UDC 681.327, publ. 64.08.08, No. 32).

Closest to the claimed technical solution is a patented method for application No. 115462 (prototype). The method includes converting cifrovoy shape of the detected image and the reference image, the coordinates of the detected and reference images, the comparison and determination of the coincidence of the detected path and the reference path. For example, recognition and reproduction of completed forms ABBYY Fine Reader 6.0 is designed for input to the computer using a scanner texts, completed questionnaires, registration cards, etc. the System provides recognition a few images of characters printed text by comparing with the reference signs of the computer catalogue of typefaces and translates these text documents in electronic form - editable files, usually format (DOC) text layout, in which the necessary amount of rows is achieved by a uniform increase in word and letterspacing distances (apresa). After scanning, the system recognizes the received graphemes characters and highlights all the questionable characters. Image with such markings are sent for verification to match with the original spelling and syntax of the text.

This method of image recognition, printed with a wide range of quality printing, where the spatial spectrum of interference in the form of spots and dots overlaps with the spectrum of the image, where the text characters have nephropathy, dull and blurry images, defects in the form of merging pollution the different elements of the parallel strokes, broken lines and uneven thickness, and in addition, the skew lines do not provide the necessary accuracy of recognition, as the percentage of characters that has some distortion is large enough. As a result, all the questionable characters of text ambiguously classified, with large errors, is poorly understood and require a lot of time for the verification process. Furthermore, the method does not ensure the desired facsimiles when playing a geometric layout of the text. When recognition can take place using other similar typeface, ograblenie General contour of the graphemes of the character font, the violation ratio in thickness of the main strokes of a character in relation to the original and set the layout of the text stream. There has been a breach in the length of the phrases of text, word lengths and the values of aproche, push the justification of the text in relation to the original, which is very important, because from the compositional integrity of the text depends on its facsimile visual perception.

The main objective of the present invention is to ensure the reliability of character recognition of text printed products and achieving facsimile reproduction points of the printed characters of any font, garnitura-pins, and inlaid layout in a wide range of print quality.

Positive achiev that is achieved by the determination of the coordinates of the characteristic line of the contour of the detected and reference images, comparison and definition match each of the detected images of the characters of the font of the original reference marks (marks cryptonites), with the same font and font size, produce extra on the area of their printed points optical overlay and uniform combination of positive (in direct contrast, the reference image corresponding to the negative (reverse contrast) a discernible image. The coordinates of the characteristic line of the contour of the detected image is produced using the corresponding reference image by finding the location of the optical center of the font on the original alignment and uniform overlap area of the printing points of the eponymous characters. As a reference the font when performing the comparison operation using the headset characters recognizable image, optimized for quality printing points, size and width to ensure the unity of their clothing construction and creation of electronic cryptonites.

The comparison of the detected and reference images produced by prior interactive markup recognizable image of text about what Iginla by defining and finding the axial lines of the contours of the characters and highlight them left and right edges of the strip type, the upper and lower lines of the font and leading lines.

Thus, verification of the character recognized text printed products optical overlay and combining the eponymous characters of a text and determining location coordinates of the optical center of the font on the original, increases the reliability of character recognition of text printed products and provides facsimiles play points and layout of printed characters of any font, garnitura-pins, and inlaid layout of the text.

The possibility of carrying out the invention is illustrated presents a block diagram of a method of facsimile recognition and text reproduction of printed products.

Method facsimile recognition and text reproduction of printed products includes two stages: the stage of creation and facsimile reproduction cryptonites (standard) on the basis of graphic and metric analysis of the font and the stage of recognition and reproduction of the text of the printed products. The second stage includes two levels of recognition. The first level is the recognition by comparison with samples of all classes of the font of the reproduced image of the text with ABBYY Fine Reader 6.0, and the second software methods and method of interactive graphics is a more accurate comparison and definition matches graficas them and metric characteristics of the font settings and features inlaid layout detected optical image overlapping and sequential or selective alignment with the standards, the set of characters which are recognized facsimile images of characters of the original.

In the first stage, spend graphic and metric analysis of the characters of the font and determine the method of printing of the original text. Enter the image text in the computer using a scanner with a resolution of 1200 dpi in color mode RGB Color 24 bit per pixel and translate the text the image of the document in electronic form.

Next on the monitor screen in an enlarged scale up to 50 times on the printed image of the printed text of the form choose the most precise printing point alphabetic characters and other graphic symbols used in the text. Based on the graphic and metric analysis to determine the height and width of the eponymous characters, the thickness of the main and auxiliary lines and their correlation. The set of characters used to create electronic cryptonites (reference).

The hardware and software level processing provides the functions of quality input images and useful information. In Photoshop 4.0 use algorithms contours allocation of marks, i.e. eliminate typical graphical distortion in the sample characters of the alphabet, due to the printing process in one way or another. Produce a tone correction of images, text, enhancing the brightness and contrast, removal of the veil is and by sharpening the edges of the images of the electronic filters. The operation of the selection circuits, the proofs of graphemes characters, detection and recovery points marks on the image of the impression from the printing plate is performed on the monitor screen with zoom mark up to 50 times compared to the original. Methods electronic retouching remove rostick around the printed points, restore blank spaces in the grapheme letters, maintain the character of the cuttings, and the thickness ratio of the main and auxiliary strokes, provide and maintain thereby the unity of the clothing build a font. Using editing programs fonts (FontLab) import the recovered font and exports it in the TrueType format, and then as new font injected into the directory standards recognition program ABBYY Fine Reader.

After scanning and processing the digital form of the original text recognition system and imaging ABBYY Fine Reader 6.0 using sampling headset characters of reference (cryptonites), verification (comparison with the original on the screen and read text) and check the syntax and spelling write text into an editable file that is exported with a resolution of 1200 dpi in the package of vector graphics CorelDraw 10.0, allows you to convert text to graphics and edit them interactively. nteractive editing text you can select individual characters of a font to change these settings: angle, height, indent, spacing between characters and between words, to change up the layout of the text.

Method e-transformation in processing package graphics CorelDraw 10.0, images of the original text and characters of cryptonites (standard) lead to a single scale.

In the selected format and scale with the required resolution label original, in the package of vector graphics CorelDraw 10.0, which resulted in the use of guidelines Guidelines on the original define and distinguish the left and right sides of the strip type, the lower and upper lines of the font, the line spacing of lines by identifying and highlighting centerline thickness contours characters using simulated for these purposes, rectangular samples with a height of one of the rectangles of equal size the font and the height of another, equal to the line spacing (leading lines).

Simultaneously with the detected image in the input file import a digital image of the original. CorelDraw allows you to project on the screen of monitor multiple images simultaneously to invert and to perform addition, subtraction (positive-negative) of these images. The same combination of images produced by the optical overlay and adding this the traditional signs of cryptonites in direct contrast with the corresponding detected signs of the original in the reverse contrast methods of vector graphics package CorelDraw 10.0. This allows you to expand the capabilities of the verification process by further comparing the detected image with a reference sequence or selectively and implement additional control matches (facsimiles) on the graph of each of the detected images of the characters of a font with the same reference signs in the area of their printed points in the scale up to 30 times.

Along selected and marked the top and bottom lines of the font in control of geometrical parameters of the dial layout of the text: the location of the combined point font, their mutual position relative to each other, the correctness of text word wrapping, justification, and line spacing of the lines, the location of the band typing in General. In case of deviation from the original and interactive way programmatically perform the change of sign of cryptonites (standard), its reversal in accordance with the same image of the original. Achieve 100% overlap area of the printed points serial optical overlay and combining the eponymous characters (positive-negative). Then fix the degree of coincidence of the images of the two eponymous characters or whole words across the strip set and only when they are full coincidence with the original software by fixing the identity (f is similiraty), i.e. the location of the image of a character or group of characters of text cryptonites (standard) on the runway of a text set of the original and copy their code into the output file with the corresponding coordinates X, Y, angle β the location of the font.

On a laser imagesetter resolution X dpi receive text film Assembly for the manufacture of printing plates to the printing method corresponding to the original.

Using the proposed method provides, in comparison with existing methods, the possibility Fax recognition and text reproduction of printed products to small formats, printed in a wide range of print quality and the degree of aging of the publication. When this is played accurately used the typeface print at the expense of restore points printed characters of any font, garnitura-pins, while preserving the metric and the graphics settings, characters, complex layout elements, length of phrases, transfer and justification of lines that can be used in the restoration and re-release old rare specimens job printed products to save composite-integrated text and provide facsimile visual perception.

Sources of information

1. Patent (19) EPO (EP), (11) application No. 0114248, CL MKI G 06 To 9/78, UDC 681.32, publ. 84.08.01 No. 31, (Academy of Sciences of the log).

2. Patent (19) EPO (EP), (11) application No. 115462, (51) MCI G 06 To 9/30, (53) UDC 681.327, publ. 64.08.08 No. 32, (prototype).

3. Gorelik A.P., Skrypnik V.A. Methods of recognition. - M.: Higher school, 1977.

4. Rahimova E.A. Control the size and shape of the images in photoreproduction processes. - RG Printing industry. Issue 1, Moscow, "Book", 1975. 40 C.

5. Witt A.A. Determination of optimum process parameters decentralized production of Newspapers using facsimile transmission. Abstract of PhD thesis. - M., 1971.

1. Method facsimile recognition and text reproduction of printed products, including the digitization of the detected and reference images, digital processing by determining the coordinates, the comparison and determination of the coincidence of the detected and reference paths, characterized in that the coordinates of the characteristic line of the contour of the detected image of the symbol produced by using the corresponding reference graphic images by finding the values of X, Y coordinates, angle β location of the optical center of the characters superimposed on the area of the printed points digital image - in direct contrast reference to the corresponding discernible in the reverse contrast.

2. The facsimile way recognize what avania and text reproduction of printed products according to claim 1, characterized in that the reference font when performing comparison operations using the sample characters recognizable image that is optimized for printing point.