DE19943156C2 - Process for increasing print quality - Google Patents

Description

The demands on print quality grow with the increasing use of printers and scanners. The Printing should be both quick and clear. The most most commonly used halftone processes in conventional Printers are shown below.

1. Controlled blur

The image in a given area becomes point by point encoded to visually simulate grayscale d. H. the Grayscale is reduced in price by the resolution increased.

2. Error diffusion

The grayscale error of pixels along one given direction is used and the next pixel is determined by a certain accumulated error ratio black or white pixel set.

The above methods are well suited for graphics rendering, however, do not result in satisfactory results Print quality for text. To overcome this problem was devoted a lot of effort to his research, see for example U.S. Patents No. 4411015, No. 4503556, No. 4504972, No. 4513442, No. 5091964, No. 5280367, No. 5335290, No. 5375197 and No. 5381241. A division is made according to these patents between text and graphic blocks by adding pixels according to the similarity or combination of gray values can be combined and separated. Assessing a divided blocks as text or graphics also happens through his histogram or shape. For a correct one However, splitting needs a significant amount of data to Example the data of the whole page, taking into account what for a fax machine or a copier that prints in real time should be unsuitable.

In addition, the problem mentioned at the beginning be overcome by using the previously processed ones statistical data set the edge threshold and the Edge subdivision is enforced; however this leads to blurred or broken edges in the printed image.  

From German Offenlegungsschrift DE 197 19 378 A1 a method of enhancing and printing text in an image known. In the process, text is based on the characteristic identified that he alternated left and right Edges. Then the print quality is only in the areas identified as text improved.

An edge detection method is known from DE 689 27 696 T2 known in which the difference in pixel density between a Target pixels and neighboring pixels with a threshold is compared and an edge is identified if the Difference exceeds the threshold. The threshold is included variable depending on the pixel density.

DE 38 38 730 A1 describes an image processing method Known in the image data to be printed on a printer with the error diffusion process are processed.

In Bässmann H. et al .: "Contour-oriented processes in the digital image processing ", Berlin, Springer-Verlag, 1989, Pages 10 to 20, edge detection methods are described, where gradient operators are used to make an edge taking into account their edge direction.

It is an object of the invention to improve the print quality of text to improve without sacrificing the print quality of graphics influence.

The goal is achieved through a procedure with the Features according to the independent claim.  

The invention provides a method that Print quality of grayscale or RBG color tones Text increases without the print quality of graphical representations to influence.

The invention also provides a method for increasing the print quality with adaptive edge threshold, where the edge feature of a just processed Pixel row is used to determine the edge threshold of the to change the next row of pixels, thereby printing in Real time is achieved.

Furthermore, the invention provides an algorithm for Select a preferred edge feature from several Edge features specified to indicate a difference in direction the operation of the edge operator and an edge roughness to remove the error diffusion.

The various goals and advantages of the present Invention will become more apparent from the following detailed description in conjunction with the enclosed Drawing figures showing the following:

Fig. 1 is a flowchart of a conventional color printing process.

The flowchart Fig. 2 according to the present invention.

Fig. 3 shows the RGB and CMY conversion as well as the edge detection.

Figure 4 is the flow diagram of the edge threshold calculation.

Fig. 5 is the flow diagram of the edge assignment.

Fig. 6 is the flowchart of the print mode conversion.

Figure 7 is the flow chart of the CMY and CMYK conversions.

Fig. 8 shows the gray level changes of a scanned image.

FIG. 9 shows the 3 × 3 edge operator detection for the image according to FIG. 8.

Fig. 10 shows the steps of the error diffusion along the edge feature.

Fig. 11 shows the difference after the error diffusion, which increases the print quality.

Figures 12A and 12B illustrate the effect of the invention.

The invention aims to provide a method which improved printing of text in real time without that To influence print quality of graphical representations, so that the process is advantageous for printing, faxing, copying and Image processing can be used.

Fig. 1 relates as an example to the case of color copying; As shown in the figure, the document is first scanned to generate digital RGB data, and the RGB data is subjected to CMYK transformation, halftone processing, and graphics and text enhancing processing to produce a CMYK halftone image produce; finally the CMYK halftone image is sent to the printer for printing. For simple color printers, the resolution is 600 dpi and the CMYK point is divided into the class of the points to be printed or the class of the points not to be printed. Therefore, the scanned document needs a text-enhancing treatment in order to limit the text printing without influencing the image printing.

The general document can fall into two categories be classified. The first type of document has obvious edges, such as text, lines, symbols, with abrupt color and grayscale. The second type of document has no obvious edges, such as pictures or image data with varied color variations and grayscale.

According to the above difference between the two types of documents, the text part of a document can be separated from the image part of a document due to the presence of edges. Fig. 2 shows a preferred embodiment of the invention in the case of application to a color image. The flow chart according to the invention for an RGB color image runs as follows:

1. CMY transformation

The RGB image should be subjected to a CMY transformation before printing. Printing actually works in CMYK mode, but the K color uses parts of the other three colors (CMY). This makes edge detection difficult. In the present invention, the method begins with a simple CMY transformation, where f (x) is the gray level of the pixel x:

f _C (x) f _R (x)

f _M (x) = 255 - f _G (x)

f _Y (x) f _B (x)

2. Edge threshold

In general, an edge threshold is required to achieve this To judge the presence of an edge, taking the default value the edge threshold of past experiences or of a statistical evaluation of the entire document page depends. This makes real-time printing difficult. In the present invention becomes the edge feature of the straight processed series of pixels used to Change the edge threshold of the next row of pixels, which enables real-time printing.

First, the initial edge threshold is determined from past experience, and a 3 × 3 edge operator is used to find the edge feature, including pixel coordinates, edge direction, and edge feature value. If the edge feature value is greater than the initial edge threshold value, the averaged edge feature value is weighted. After a row of pixels has been processed, the edge threshold is changed by the weighted average. If the edge threshold is less than the initial edge threshold, then the edge threshold is set as the initial edge threshold. The flow chart is shown in FIG. 4.

3. Edge detection and distribution

In the present invention, a 3 × 3 edge operator used to capture edge features. In particular the present invention uses the directional feature of Edge for edge distribution. The present invention uses two data bytes for recording features of a pixel, including the location of a black pixel, one Edge change coefficient, a directional feature of  black and white pixels.

If the edge feature value is greater than the edge threshold value, the direction of the maximum edge feature value is first found and the maximum edge feature value is encoded according to the pixel distribution principle in the middle of the pixel. The center pixel of the edge operator is recorded and the information originally assigned to the center pixel is deleted. The directional distribution for further pixels then takes place, and the edge features are assigned to the pixels, as shown in FIG. 5.

4. Distribution of the directional characteristic

For a general picture, the gray scale increases when a dark pixel on the left and a light pixel on the right located. For a realistic picture, according to the Edge processing multiple edge pixels with one above the Edge characteristic value lying edge value can be obtained. For the print quality it is important like the black and the white pixels can be located at the correction point. The present invention proposes an algorithm that a preferred edge feature from multiple edge feature values selects to eliminate the difference in direction caused by the work order of the operator and the edge roughness is caused.

The proposed algorithm uses the guideline each pixel to an AND (&) or OR (¦) operation. If for example, the darker edge of an image Pixel on the left and the lighter pixel on the right and if after an edge operation from top to bottom and left to the right is the edge feature on the left side, checks the inventive method, whether the left pixel of the Edge operator the characteristic of a white left-hand side Has pixel. If so, it means that the left pixel is determined as a white pixel. Since the Pixel in the direction is determined as a white pixel, the following center pixel is determined to be the Has the feature "white pixel on the left side".  

If the answer to the above test is "not applicable" is in the left pixel or the center pixel of the edge operator through the interpolation and Error diffusion method the feature "black pixel on the left side "inscribed.

If the feature "black Pixel on the left side ", should the feature "black pixel on the left side" be deleted at the left pixel to prevent the feature "black pixel on the left side" is repeated. In addition, at the center pixel, that Feature "white pixel on the left side" also to be deleted. Then the characteristic "black pixel written on the left side "in the center pixel, and the feature "white pixel on the left" is used in the right pixel written.

If the feature "black pixel on the left side" are written in the left pixel of the edge operator the left pixel remains unchanged if it does Characteristic "black pixel on the left side" already has. Otherwise, the feature "black pixel on the left side "written in the left pixel, and that Feature "white pixel on the left" is in the Center pixel written.

When a new edge feature occurs and the directive is not left, the feature distribution is the same, but that Procedure is different.

5. Interpolation and error diffusion

The data obtained by scanning generally has a resolution of 300 dpi, and the normal print resolution is 600 dpi. The 300 dpi image is in a before printing 600 dpi CMYK image converted. However, the conversion process time consuming and requires a lot of storage space, and - what am Most serious is - the edge feature is lost. In the present invention becomes an interpolation method used to linearly transform the pixel of the 3 × 3 edge operator interpolate a 6 × 6 pixel. In addition, a  Error diffusion techniques are used to determine whether that Feature "black pixel in direction r" in which the Direction r concerned pixel or the center pixel of the Edge operator should be written, and around the Determine the position distribution of the black pixel. at the error diffusion method is the gray scale after Interpolation either print (grayscale = 0) or not Print (grayscale = 255), the error value is the same Pressure value minus the gray level after interpolation, and the Error value is on the pixel of the next Interpolation point with a coefficient of expansion (t) transmitted when the grayscale error does not differ from a 127 value exceeding a value above 127 accumulated or vice versa, lies in the interpolation pixel an edge in front. By using this procedure Characteristic "black pixel in direction r" can be determined in order it in the pixel concerned with the direction r or the To write the center pixel of the edge operator.

As for the local distribution of the black pixel, it is checked whether the position of the black pixel for the pixel relating to the direction r or the Center pixel with the directive through the Error diffusion principle. In general, the Position of the black pixel with the corresponding directive matched the change in position of the black pixel should occur, even if the Grayscale accumulation is insufficient, causing a Edge shift is prevented. Therefore, the Pixel position more realistic with the grayscale of the real image to be matched.

6. Edge modification

Because of the different resolution between the The original image and the printed image can be along the edge a difference of one pixel in the same direction a low resolution image, but a difference of two or have more pixels in a high-resolution image, therefore the edge of the picture appears interrupted. The present  Invention uses two rows of data with one Directional feature to solve this problem. The diagonally neighboring data is checked to see if it has the same directive with respect to a black pixel. If so the edge pixel is not continuous, and one of the diagonally adjacent information needs to be changed to to represent a continuous edge.

7. Print mode conversion

If the print mode is changed, it should Directive feature of the edge pixel to be shifted to the Increase print quality. The 300 dpi RGB image should be one be subjected to appropriate conversion to a 600 dpi To achieve pressure effect. Therefore the direction of the Edge pixels are converted to the 600 dpi effect display. Generally, each pixel has eight Neighboring pixels, in other words eight directions. in the normal print mode creates one dot of the original image four Positions under normal print mode. That is why the data from eight neighboring pixels to those four printing positions be distributed. The present invention uses one byte to code the possible position assignment under the normal print mode including mapping black and white pixels as well as the assignment of black and white pixels for edge modification.

In other words, at the dots of the original image, each edge pixel has two bytes of data for recording the edge feature, and these two bytes of feature data correspond to one byte of print data according to the difference in the printing mode, and the flowchart is shown in Fig. 6.

8. CMYK conversion

The result of the above steps relates to the edge pixel distribution in the CMY print mode, but the actual print mode is the CMYK mode. Therefore, the edge pixel distribution of the CMY mode must be converted in order to reduce the color ink consumption of the color printer and to prevent the overlapping error of the CMY colors. The steps are as follows. If a pixel has an edge feature and a CMY color is to be printed in print mode, then the CMY color is replaced by a K color for printing; if a pixel has an edge feature and no CMY colors are to be printed, then the K color is not printed either; if the pixel has no edge feature, the printing condition is not changed; the flow chart is shown in FIG. 7.

A preferred embodiment will be described below with reference to FIGS. 8 to 11. The image is assumed to be a monochromatic grayscale image and a scanned 300 dpi image has the grayscale values shown in Fig. 8 and a 3 × 3 edge operator is used. The gray levels shown in Figure 8 drop from 255 to 0, indicating the presence of an edge. Fig. 9 shows the method; provided that the three operators shown in FIG. 9 generate an edge feature value above the edge threshold value, a preferred edge should be found among these results. Fig. 10 shows the method by assuming that the diffusion coefficient is 0.5.

In the first step it is known that the edge feature value of the operator 1 is greater than the edge threshold value, the pixel distribution coding obtained from the largest edge feature value is recorded in the pixel C, and then an interpolation to the edge operator 1 takes place. If the gray level b 'is not to be printed after interpolation (gray level 255 ), the error ( 255 -b') multiplied by the error coefficient 0.5 is diffused to gray level c. As shown in this figure, the accumulated gray level error in gray level c is above 127, indicating a whitish color; if the gray level of c is not to be printed (gray level 255 ), the error multiplied by the error coefficient 0.5 ( 255 -c ') is diffused to gray level d. It is found that the gray level d is below 127 (blackish). In other words, the gray level changes from white to black during the transition from pixel C to pixel D. Therefore there is an edge. A recording of the white pixel on the right is stored in pixel C, and a recording of the black pixel on the right is stored in pixel D.

The error diffusion method is also used to determine whether the edge direction of the pixel D needs to be changed. That is, it is assumed that the gray level c 'is not printed. The error multiplied by the error coefficient 0.5 ( 255 -c ') is diffused to gray level d, it is found that the accumulated gray level error at gray level d is below 127 (blackish), the black pixel of the edge becomes in the direction of the gray level d and not assigned in the direction of grayscale d '.

In the second step, the operator 2 is carried out after the operation of the edge operator 1 . It is found that the edge feature of operator 2 is above the edge threshold, which means that there is an edge. First, the pixel assignment coding, which is obtained in relation to the maximum edge feature value, is recorded in pixel D, then the interpolation takes place on the edge operator 2 using the error diffusion principle. The result is that at the transition from pixel C to pixel D, the gray level changes from white to black. However, pixels C and D have a record of the edge feature, the record of which is a. Modification with a logical operation requires, that is, a new directional feature is increased and this feature is retained. Finally, the error diffusion method is used again to determine whether the edge assignment direction needs modification, the result is the same as in the first step.

The edge operator 3 is carried out after the end of the second edge operator 2 . It is found that the edge feature of operator 3 is above the edge threshold, which means that there is an edge. First, the pixel assignment coding, which is obtained in relation to the maximum edge feature value, is recorded in the pixel E. The pixel white edge direction (pixel D) is checked. If the pixel has a record of a black pixel on the right, indicating the presence of an edge, no modification is required.

Combining the above steps, the pixel C records the information "white pixel on the right side", and the pixel D records the information "black pixel on the right side", and the direction is changed to the left. Assuming that the above 300 dpi edge image has the grayscale distribution shown in Fig. 11a after the conventional error diffusion process and 600 dpi printing, the result is shown in Fig. 11b, where W "do not print" and B "print "means. As shown in this figure, the edge does not appear. Figure 11c shows the edge feature according to the method of the present invention, the record in pixel C with the white pixel on the right is encoded as "do not print", and the record in pixel D is "black pixel on the right", and the direction is changed in the left direction. As shown in Figure 11c, the edge is smoother and more visible.

FIG. 12A and 12B show the effect of the invention. Fig. 12A shows the printing result of a scanned image without processing by the present invention, and Fig. 12B shows the printing result of a scanned image with processing by the present invention. As can be seen from these figures, the edge of the print result is smoother and clearer by the present invention.

The present invention has been described with reference to FIG described their preferred embodiment, but understands it that the invention is not limited to the details of Embodiment is limited. Different replacements and modifications are described by the above description proposed, and others arise for relevant Professionals.

Claims (8)

1. A method for increasing the print quality of grayscale and color images that are not available as CMYK images, with the following steps:
CMYK halftone

• the document to be printed is converted into a CMYK image according to the following steps A to D, and halftone processing is carried out for each color in the CMYK colors to determine the color of a pixel in the image to be printed;

A. CMY conversion

• - The document to be printed is converted into a CMY image, with a conversion of RGB gray levels (f _R (x), f _G (x) for each pixel of each of the three colors C, M and Y, f _B (x)) in CMY grayscale (f _C (x), f _M (x), f _Y (x)) is performed around the edge in the CMY colors for each one of the three colors C, M and Y, and at the same time for all three colors;

B. Edge detection

• the presence of an edge feature is determined by the gray level difference existing in the input document, no processing takes place if there is no edge feature and the edge feature is recorded if an edge feature is present; by the following process:
  (B.) image data are computed, which are enclosed by an m by n window, the edge threshold value is changed by means of an edge operator by the edge feature of the edge gray level difference, and the feature record of the edge pixel is found out from the edge operator by the edge directive;
  (C.) the edge feature value in one row exceeding an initial edge threshold is weighted and averaged to change the edge threshold and thereby provide the edge threshold of the next row;

C. Print mode conversion

• - The feature record of the edge pixel is assigned according to the print mode to determine whether each edge pixel is printed;

D. Conversion for CMYK print mode

• - The CMY edge pixel is converted into the CMYK edge pixel.

2. The method according to claim 1, wherein the edge detection step B further comprises the following substeps:
(A.) Set an initial value

• - An initial value of the edge feature is set before the sub-steps (B.) and (C.);

(D.) recording feature

• two bytes of data are used to represent the edge feature of a pixel - including the location of a black pixel, the edge modification coefficient, the direction of the black and white pixels;

(E.) Feature assignment

• if the edge feature value is greater than the initial edge threshold, it means that there is an edge within the image of the m by n window, the edge feature is recorded in a preferred pixel by an operation according to the interpolation and error diffusion method;

(F.) Edge modification

• If, when using two rows of data with directional features, two diagonally adjacent information data indicate a black pixel and the same direction, one of the two diagonally adjacent information should be changed so that the edge directions of the two diagonally adjacent data are close to the center of an adjacent point.

3. The method of claim 1, wherein the conversion for the Print mode includes the following substeps: the edge feature of the Edge pixel will correspond to the pixel in print mode assigned to the feature record, the white pixel feature becomes the first and the black pixel feature the second by means of an operation, for example an OR operation, processed to add the pixel feature in the Process print mode, the directional feature of any white Pixel corresponds to the white pixel in print mode, the Directional characteristic of each black pixel corresponds to that  black pixel in print mode. 4. The method of claim 1, wherein the CMYK conversion includes the following steps: if a pixel has an edge feature owns and should print CMY colors in print mode, then the CMY color is replaced by the K color for printing; if a pixel has an edge feature and no CMY If colors are to be printed, then the K color will not either printed. 5. The method of claim 2, wherein the interpolation Gray level of image data linearly interpolated in the m by n window, to produce grayscale pixels of 2m by 2n when the Edge feature value greater than the initial edge threshold is. 6. The method according to claim 2, wherein in the conversion of the Grayscale image data from the m by n window Interpolation in 2m by 2n grayscale pixels interpolated grayscale is black or white, depending on the Directive of the edge feature, and the error to the next interpolated pixel is accumulated when the color of the Grayscale after error accumulation from black to white passes or vice versa, and feature data in the m by n data be set up to record the edge feature. 7. The method according to claim 2, wherein - if in the m times n Data the edge feature is recorded - an OR Linking is used to record another Maintain edge feature, and an AND operation is used to record the same edge feature Clear. 8. The method according to any one of claims 1 to 7, wherein the printing document is available as an RGB image.