JPS6182577A - Picture recording device - Google Patents

Abstract

PURPOSE:To express a true half tone by classifying the input picture to three kinds of a character/a line picture, a network point/half tone picture and other picture, and executing the dither processing suitable to the picture classification for every these classifications. CONSTITUTION:A picture quality improving circuit 14 compares and decides an assembling of a picture information pattern obtained by making binary a Laplacean of the inputted picture information, with a standard pattern pre pared beforehand, further, generally decides the size of Laplacean and the assem bling, and classifies an input picture to a character/line picture, a network point/a half tone picture and other picture. After part of Laplacean is added to picture information for a character and a line picture, a binary circuit (ROM)17 dither-processes and makes binary and after picture information is made average for the network point/a half tone picture, dither-processes, and for other picture, dither-processes picture information directly, or after part of Laplacean is added to picture information, the circuit dither-processes. The signal, which is dither-processed, is supplied to a printer 3.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an image recording apparatus that performs dither processing on an input image and records and outputs the dithered image as a binary image consisting of a dot pattern.

[Technical background of the invention and its problems]

, a! Halftone images with light (gradation) and 21i! When displaying (recording) images using a binary image display (recording) device such as a dot printer, a dither method is often used as image processing. However, when image processing is performed using systematic dithering, etc., although it is possible to effectively record and output the inside lecture image, the resolution of text and line drawing components included in the input image is significantly reduced, and the recorded and output image It had some drawbacks, such as being very unsightly.

Conventionally, in order to solve the above-mentioned problem, for example, the maximum value and minimum value of the image signal are determined for each specific block of the input image, and when each of these values exceeds a certain value that constitutes one city, the value of the specific block is determined. It has been proposed to perform fixed 21ili conversion when determining that the image is a line image, and to perform dithering in other cases.

Based on a similar idea, it has also been proposed to obtain the Laplacian of an input image and determine whether or not the input image is a line image from the size of the Laplacian.

However, in such image type determination, for example, 2
When a halftone image or the like expressed in halftones using 1m data is input, it is simply determined as a binary image. As a result, the input image is subjected to fixed binarization processing and then output as an image, resulting in a problem that halftone information of the input image is lost.

[Purpose of the invention]

The present invention was made in consideration of such circumstances,
The purpose is to accurately classify and judge the type of input image, including halftone images, etc., and to perform image processing appropriate for each image type to improve the image quality of the output image. An object of the present invention is to provide an image recording device.

[Summary of the invention]

The present invention converts input images into text/line drawing boxes, halftone/halftone images,
For example, a combination of image information patterns formed by binarizing the Laplacian of image information read and inputted by photoelectrically converting an image original, and a standard pattern prepared in advance. Furthermore, the size of the Laplacian of the above image information and the combination of the Laplacian are comprehensively judged to classify the image type of the input image. For this purpose, after adding a part of the Laplacian to the above image information, it is dithered or fixed binarized, and the dither size at this time is Ml), and
For halftone/halftone images, the above image information is averaged and then dithered, and the dither size at this time is set to M2), and for other images, the above image information is directly dithered. Alternatively, after adding a part of the Laplacian to the image information, dithering is performed (the dither size at this time is M3), and the input image is subjected to different image processing for each image type. Record and output. In addition, at this time, the dither size M1 described above is used. M2. M 3,
For example, the relationship M1≦M3≦M2 is established.

〔Effect of the invention〕

Thus, according to the present invention, input images are classified into three types of images consisting of text/line images, halftone/halftone images, and other images, and for each of these image types, a Since the dither processing is performed, the halftone image is not determined to be a binary image as in the conventional method, and its faithful halftone expression can be achieved.

In addition, as mentioned above, the image type can be determined not only from the high-frequency response of the image, but also from a combination of patterns obtained by binarizing the Laplacian, etc., so that each feature of the line image or halftone image can be well captured. Since the type of input image can be classified and determined with high accuracy. In addition, by varying and adapting the dither of the optimal matrix size to express the image based on the determined result, the conflicting properties of resolution and number of gradations peculiar to dithered images can be resolved depending on the type of image. This allows for optimal matching.

Therefore, it is possible to comprehensively improve the image quality of the image to be recorded and output, and it is possible to obtain an image that well reflects the information contained in the input image, which brings about great practical effects.

[Embodiments of the invention]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a schematic diagram of an embodiment of a thermal transfer color printer to which the present invention is applied. This apparatus reads and inputs a color image original 2 using a color scanner 1, processes the input image signal, and then outputs it to a thermal transfer color printer 3 to produce a color copy image.

A black reference plate 5 and a white reference plate 6 are provided at the side positions of the glass table 4 on which a color image original 2 to be subjected to color image copying is placed.
When the color image original 2 is read by , the color information is read as a black reference level and a white reference level, respectively.

The color scanner 1 includes an illumination light source 7 such as a fluorescent lamp that illuminates the color image original 2 and includes each of the reference numbers 5.6, and
The color image original 2 illuminated by this illumination light source
The reflected light of C is detected via the rod array lens 8.
A CD color line sensor 9, an amplifier 10 for amplifying the image signal read and input by the CCD color line sensor 9, and an A/D converter 11 for digitally converting the image signal obtained via the amplifier 10. Prepared and configured. The CCD color line sensor 9 is composed of, for example, red (R), green (G), and blue (B) color filters arranged dot-sequentially. Each color component signal over the scanning line is obtained point-sequentially. Such point-sequential color component signals covering one scanning line are sequentially outputted from the CCD color line sensor 9 and sent to the A/D converter 1.
1, each is digitally converted and output.

The color scanner 1 is moved and scanned in the six directions of the arrows in the figure, and first scans the black M semi-plate 5 and the white reference plate 6 to obtain a black reference signal and a white reference signal, and then scans the color image original 2. The image information is read by scanning sequentially. In this way, each color component signal of the image signal which has been color separated and read and input through the CCD color line sensor 9 is given to a standardization circuit 12.

The standardization circuit 12 first sets a black reference level and a white reference level for the image signal read from the color image original 2 based on the black reference signal and the white reference signal, which are input after each of the reference signals. For each color component signal of the image signal, the white reference level is '''0°°, and the black reference level is I.
Correction parameters for standardization are set as I111.

Thereafter, the color image original 2 is sequentially scanned by the color scanner 1, so that the image signal of the color image original 2 is converted into the color-separated color component signals R, G, and B.
When inputted dot-sequentially for each scanning line, the normalization circuit 12
The color component signals R, G, and B are each standardized in accordance with the correction parameters as described above. The matrix circuit 13 receives each of the standardized color component signals R.
,G.

From B, its luminance signal component I and color difference signal component C1°C2
This is what we seek. The image quality improvement circuit 14 which receives an input image signal consisting of such a luminance signal component (2) and a color difference signal component (1. The type of the input image is determined based on the information obtained by binarizing the Laplacian of the input image signal.The color conversion process, binarization process (dithering process), etc. described below are performed according to this judgment result. Each is controlled according to the image type.

The color conversion circuit 15 converts the respective color signal components I and CI.

From C2, each color separation signal corresponding to the print color of the thermal transfer printer 3 is sent, for example, yellow (Y).

Magenta (M), cyan (C), and black (K) are obtained in accordance with the amount of ink to be printed. This color conversion process is basically performed according to Neugebauer's equation, masking equation, etc., and for example, the color conversion process is performed using a ROM in which the values of each color separation signal calculated in advance according to these equations are tabulated.

The switch circuit 16 selects the color separation signals Y, M, C, and K obtained through the color conversion circuit 15 in accordance with the color ink ribbon set in the thermal transfer printer 3. The ribbons are yellow (Y), magenta (M), and cyan (C).

Assuming that black (K) is selected in sequence, the color separation signal Y is first selected in accordance with this selection order. The selected color separation signal Y is then subjected to a predetermined dithering process via the binarization circuit 17 and then provided to the printer circuit 18. The binarization circuit 11 is
In order to display a halftone image using a printer device capable of only 216 recording, such as the thermal transfer printer 3, data is converted by, for example, a systematic dither method. The printer circuit 18 drives the thermal head 19 of the thermal transfer printer 3 in accordance with the color separation signal Y subjected to such data conversion, and transfers and records the ink of the yellow color ink ribbon onto the recording paper 20. There is. Note that this printer circuit 18 corrects changes in characteristics due to temperature of the thermal head 19, etc., and always transfers ink at a constant temperature of 19 degrees, that is, the above-mentioned 2 ft! [Image recording of a constant density is performed for data.

The above processing is performed in real time in synchronization with the reading and scanning of the color image original 2. And this color image original 2
During the first scan, only the color separation signal Y is applied to the printer circuit 18 as described above, and a yellow color separation image is transferred and recorded on the recording paper 2o.

After recording the yellow color separation image in this way, the color ink ribbon of the thermal transfer printer 3 is switched from the yellow color to the magenta color, and the reading scan of the color image original @2 is repeated in the same manner. be exposed. In this case as well, after each color component signal is standardized, these color component signals are processed to produce the color separation signals Y and M.

C and K are calculated respectively. However, since the color ink ribbon has been switched to magenta at the time of scanning the document for the second time, a color separation signal M is selected corresponding to this color, and a signal obtained by dithering this color separation signal M is used to direct the thermal head 19 is driven, and a magenta color-separated image is recorded on the recording paper 20 so as to be superimposed on the yellow color-separated image.

Thereafter, in the same manner, during the third scan of the document, the color ink ribbon is switched to cyan, color separation signal C is selected, and a cyan color separation image is recorded superimposed on each color separation image. Furthermore, when scanning the document for the fourth time, a black color ink ribbon is used, color separation signal K is selected, and a black color separation image is recorded superimposed on each of the color separation images.

By scanning the color image original 2 four times as described above, each color separated image of yellow, magenta, cyan, and black is sequentially recorded on the recording paper 20, and the color copy image 19 is recorded here. 19.

By the way, the image quality improvement circuit 14 is configured as shown in FIG. 2, for example.

This image quality improvement circuit 14 performs signal processing on the luminance signal 1 that has been processed for each color in the matrix circuit 13 in order to improve the image quality of the output display image and enable highly natural gradation expression. be.

Note that predetermined image signal processing is performed on the color difference signals CI and C2 regardless of the input image type.

That is, this image quality improvement circuit 14 includes a block 21 that measures the Laplacian of the luminance signal I, and a block 22 that binarizes the Laplacian value calculated in this block 21 and determines and detects the combination of the 21i pattern. and a block 23 for encoding the value (size) of the Laplacian, and whether the input image formed by the luminance signal I from each output of each of these blocks 22 and 23 represents a character/line drawing, or Block (ROM) that determines whether it is a halftone/halftone image or another image
24, and a block 25 that records the upper bit data of this determination result and feeds it back to the block 24.
, and then performs a first process of adding the Laplacian to the original image signal to correct high frequencies, a second process of leaving the original image signal as it is, and a moving average addition of the original image signal, according to the above determination result. A block 26 selectively controls the third process for removing high-frequency noise by removing high-frequency noise, and a block 27 controls the image processing in accordance with externally specified image type information. This block 27 instructs the self-help judgment of the image type for the input image, changes the weighting coefficient of the automatic judgment, and also forces the image processing, such as fixed binarization, character/line drawing emphasis, printed matter. This function selectively instructs general processing for images, emphasis processing for photographs, etc. Such an instruction is, of course, given by operating a selection switch or the like, according to the operator's intention.

The block 21 that calculates the Laplacian of the luminance signal r is
As shown in detail in FIG. 3, the input luminance signal I is input to line memories 32a and 32b for each line via a multiplexer 31. An adder 33...
34. However, the addition data of this lever is 3
input to the stage shift registers 35a and 35J35c,
The above addition data obtained at three consecutive timings obtained as the outputs of the shift registers 35a, 35b, and 35c are added via adders 36 and 37. Through these addition processes, addition data for (3×3) pixels as shown in FIG. 4 of the input image (i'i degree) signal is obtained as one shift of the moving addition.

The ROM 38 outputs a signal value obtained by 1/9I8 of the above-mentioned moving addition value, and by subtracting the value of the central pixel 122 of the above-mentioned (3×3) pixels from this signal using a subtracter 39, the Laplacian value is obtained. I'm looking for it. Incidentally, the output of the above-mentioned RO to 138 is obtained by averaging the input image information (removing high-frequency components).

However, the first shift of such Laplacian is block 2
After being input to 2 and binarized, specifically, being converted to 2IiII by extracting the value of the sign bit,
Line memories 42a, 42 via multiplexer 41
One line is recorded in b and 42c. Each of these pieces of information is stored in the line memory 42a.

3 pixels are read out from 42b and 42c (3×3)
The information is input to the ROM 43 as pixel information. This ROM
43, pre-stored information and the line memory 42a,
Specifically, by comparing the information read from 42b and 42c, the line memory 42a.

The information read from 42b and 42c is used as an address, and the content is determined whether the input image information is a character/line drawing, a halftone/halftone image, or another image. By looking up this table, the result of classifying the image type is output. In addition, 1g of images are output from this ROM43! The type information may be uniquely (forcibly) determined by the control information from the block 27.

On the other hand, the block 23 calculates the absolute value of the Laplacian signal and non-linearly transforms information on this absolute value. For example, a Laplacian signal represented by 8 bits of information is
It is converted into a 4-bit compressed signal. The sum of these signals for three pixels is obtained via adders 44 and 45, and the higher 4 bits of the signal are output to the ROM 24.

The ROM 24 receives each output signal of the blocks 22 and 23, that is, the result determined by the combination of binarized patterns of the Laplacian, a signal indicating 11 (magnitude) of the Laplacian, and the block 25. The image type of the input image is comprehensively determined using the destination determination information returned via the 2-bit comprehensive determination result J.
I am getting . Note that the determination information fed back through the block 25 refers to the current processing target pixel as shown in FIG.
, o, i −1, −1, i o, −1, i 1
．． -1. This is the determination result for the pixel i −1,o. Through such comprehensive determination, the type of image of the input image is classified and determined.

For example, if the input image is determined to be a character or line drawing, data J of (11) is output, and if it is determined to be an image of halftone dots, a photograph, etc., data J of (00) is output. is output. In addition, in the case of other images, (
Data J of 10) or (01) is output.

According to such a determination result, the block 26 generates the original input pixel signal, a signal obtained by adding the Laplacian signal of the input pixel (X1 degree) signal obtained in the block 21 to the original input pixel signal, or the 9-pixel average The converted signals are selectively output. Note that the amount of addition of the Laplacian is variably controlled by a coefficient unit 46.

On the other hand, in parallel with the processing for such luminance signal ■,
The color-separated color difference signals CI and C2 are averaged (low-pass processed) between the (3×3) pixels shown in FIG. 4 and output, and are processed by a filter circuit as shown in FIG. 7. The specific circuit is substantially the same as that shown in FIG. 3, and its output is similar to the output of the ROM 38.

A signal J indicating the image type determined by the image quality improvement circuit 14 is given to the 2Ia conversion circuit 11.

The binarization circuit 11 sends color separation signals Y, lvl, C, and K corresponding to the color inks to be recorded and output of the input image obtained through the color conversion circuit 15 to a switch circuit 16.
The color separation signals are selectively inputted in correspondence with the output colors of the printer 3 through the image processing apparatus 3, and the color separation signals are dithered according to the image type.

That is, the binarization circuit 11 has a pattern address controlled by a counter 47 as shown in FIG. 7, for example.
The dither pattern corresponding to the color separation signal is generated according to, for example, a 6-bit color separation signal and a 2-bit image type signal J.

The ROM is also provided with 2-bit data indicating which color component the input color separation signal belongs to. This data allows a different dither pattern (a phase-shifted pattern) to be used for each print color, but if there is no need to change the dither pattern depending on the print color, this control data is unnecessary.

The address signals X and y applied to the ROM are used to sequentially modulate the dither threshold, and are sequentially and repeatedly generated. By this address control, the color separation signal is converted into a 1-bit dither pattern signal of a predetermined matrix size and output. The basic dither matrix size is, for example, 8x8,
By switching the repetition period of the address to (8, 4, 2, 1), 8x8, 4x4, 2x2
, 1 is selectively set. Then, corresponding to each of these sizes, dither patterns as shown in FIGS. 8(a) to 8(d), for example, are selectively provided. In this case, it is advantageous to set the dither threshold values as shown in FIGS. 9(a) to 9(d) in correspondence with each of the matrix sizes in order to obtain a recorded image with high fidelity.

Binarization circuit (ROfvl) 17ha, Konoyouna 4V
! ! The color separation signal is dithered by selectively using a similar dither pattern according to the signal J.

That is, the image quality improvement circuit 14 converts the input image into characters,
If it is determined to be a line drawing, simple binarization is performed using the pattern shown in FIG. 8(d) using the dither threshold shown in FIG. 9(d).

Furthermore, when it is determined that the input image is a halftone/halftone image, the dither threshold value shown in FIG. 9 (a) is used.
) pattern. If it is determined that the input image is likely to be a character/line drawing, then the input image shown in FIG. 9 (C
) is used, and if it is determined that the input image is likely to be a halftone/halftone image, the dither threshold shown in FIG. 9(b) is used. The 8th
Each is dithered using the pattern shown in Figure (b).

That is, the dither processing is performed on the color separation signals by varying the dither matrix size and dither threshold value depending on the type of input image. The signal that has been dithered in accordance with the image type in this way is supplied to the printer 3, and the color separated image is recorded.

Thus, according to this device, the Laplacian of the input image is
By comprehensively determining the combination of digitized information patterns and the size of the Laplacian mentioned above, and classifying the image type, it is possible to perform appropriate dither processing according to the nature of the input image. Become.

For example, as shown in Figure 10, when a halftone image is scanned one-dimensionally, the read image signal is as shown in Figure (a), and when a character is scanned one-dimensionally, the read image signal is the same. The result is as shown in Figure (b). Also, please upload 1 image such as a photo.
The read image signal when dimensional scanning is performed is as shown in FIG. 13(C). Comparing these one-dimensional scanning signals, it is clear (□) that the signals of halftone images and character images have large local density changes, and therefore cannot be identified solely from the Laplacian value mentioned above. .

However, it is possible to identify an input image such as a photograph from the value of the Laplacian.

On the other hand, when examining the character images and halftone dot images, the Laplacian obtained from the character images is shown in Figure 11.
The iIn pattern and the pattern in which the Laplacian obtained from the halftone dot image is converted into 21ili are shown in FIG. 12, as shown in FIG. In other words, in the case of text and line drawings, there is generally a property that the white or black pixels are continuous, whereas in the case of halftone images, the black pixels are discrete or periodic. be. Figure 13 (a) and (b) are (3X3)
This is a histogram that shows the number of times a periodic pattern occurs when the pixel pattern data is 9-bit data, and it is possible to see that there are differences in the number of times a periodic pattern occurs depending on the input image. shown. That is, in the case of a character image, there are relatively few periodic components as shown in FIG. 13, and conversely, in the case of a halftone dot image, there are relatively few periodic components over one scanning line as shown in FIG. 13(b). Therefore, there are many periodic components. By obtaining such information from the block 22, the ROM 24 identifies characters/line drawings and halftone images that cannot be distinguished from the Laplacian values alone, and comprehensively determines the type of input image. There is.

Then, according to the judgment and identification result, if the input image is a character/line drawing, the current image signal is added to the Laplacian.
The data is added and the high-frequency components are emphasized, the signal is binarized, and this is output and displayed. In addition, when the input image signal is a halftone image, in order to express halftones, the average value of (3×3) pixels is calculated and high-frequency component noise and frequency components that cause moiré are removed. , this is dithered and output for display. Further, when the input image is not a character/line drawing or a halftone image, the original image signal is used as it is, subjected to normal dithering processing, and then displayed and output. At this time, as described above, the dither size and dither threshold are varied depending on the image type.

Therefore, the image to be recorded and output will be faithful regardless of the type of input image, and problems such as blurring of characters and line drawings and involuntary binarization of halftone dot images will be eliminated.

Note that the present invention is not limited to the embodiments described above. In the embodiment, the Laplacian was obtained as the difference between the signal average value of 3x3 pixels and the original image signal, but the average value of 4x4 pixels or 5x5 pixels may be used. Alternatively, the Laplacian may be determined by a general convolution operation. In this case, you can freely control the pass frequency band, so
It can be expected that the detection accuracy of line drawings will be improved.

Further, as a combination of binarized Laplacian patterns, a 3×3 matrix is used in the above embodiment, but a 2×4 or 3×4 matrix may be used.

Further, although color recording has been described here, the present invention can be similarly applied to black and white image recording. Furthermore,
The dip threshold and dither pattern used for dither processing may also be set according to the device specifications, and only the dither pattern size or dither threshold may be varied depending on the image type. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of drawings]

FIG. 1 is a schematic configuration diagram of an apparatus according to an embodiment of the present invention, FIG. 2 is a diagram illustrating a schematic configuration of an image quality improvement circuit in the embodiment apparatus,
Fig. 3 is a block diagram of a circuit for calculating the Laplacian of an input image, Fig. 4 is a diagram showing pixels to be subjected to Laplacian processing,
Fig. 5 is a diagram showing the corresponding pixel positions of information used in the identification judgment of the input image, Fig. 6 is a diagram showing an example of a filter circuit for color difference signals, and Fig. 7 is a diagram showing an example of the configuration of a binarization circuit. Figure 8 shows examples of dither patterns depending on the image type, Figure 9 shows dither IJlliI depending on the image type, and Figures 10 to 13 show characters/line drawings and halftone dots ii! i
FIG. 3 is a diagram showing the properties of images. ′
1...Color scanner, 2...Color image original, 3
...Thermal transfer printer, 12... Standardization circuit, 13.
... Matrix circuit, 14 ... Image quality improvement circuit, 15
...color conversion circuit, 17...binarization circuit. Applicant's representative Patent attorney Takehiko Suzue Figure 1 Figure 2 Figure 3 Figure 4 '6 ・ Figure 8 (a) (b) (c) (d) Figure 9 (a) (b) (c) (d) Figure 10 Figure 11 (a) (b) (c) Figure 12

Claims (10)

[Claims] (1) The input image is classified into at least three image types consisting of text/line images, halftone images/halftone images, and other images, and each image is added to the input image according to the classified image type. An image recording device that performs predetermined image processing for each type and records and outputs the images. (2) The input image consists of image information read and input by photoelectrically converting the image original, and the classification of the image type of the input image is based on a combination of image information patterns obtained by binarizing the Laplacian of the image information, Claim 1: The determination is made by comparing with a standard pattern prepared in advance.
The image recording device described in Section 1. (3) Image type classification is based on the Laplacian of image information
A patent claim that is made by comprehensively determining the result of comparing a combination of digitized image information patterns with a standard pattern prepared in advance, the size of the Laplacian of the image information, and the combination of the Laplacian. The image recording device according to item 1. (4) The image processing specified for each image type is processing for text/line images that adds a part of the Laplacian to the image information read and input through photoelectric conversion, and then converts it into dither or fixed binarization. Then, after averaging the above image information, processing is performed on the halftone/halftone image to be dithered, and the above image information is directly dithered, or after adding a part of the Laplacian to the image information, dithering is performed. 2. The image recording apparatus according to claim 1, further comprising processing for other images. (5) The image recording apparatus according to claim 4, wherein the amount of Laplacian added to the image information is variably controlled depending on the type of image. (6) The image recording apparatus according to claim 1, wherein the image type classification processing for the image information is performed on the luminance signal component of the image information. (7) Dithering in image processing according to the image type is performed using different dither matrix sizes, and the matrix size for character/line images is M1,
The matrix size for the halftone/halftone image is M2,
The image recording apparatus according to claim 1, wherein the image recording apparatus has the following relationship (M1<M3<M2), where M3 is a matrix size for other images. (8) The dither matrix size has a basic size of N×N (N: an integer), and the dither pattern according to the image type has a dither unit of N/K pieces (K: an integer). The image recording apparatus according to claim 7, wherein the image recording apparatus is provided as a combination of a plurality of dither patterns having different dither threshold values. (9) Dither processing according to the image type is performed by changing the gamma characteristics of dithering. G1 is the gamma characteristic gradient for character/line images, and G1 is the gamma characteristic gradient for halftone/halftone images. G2, and when the gamma characteristic gradient for other images is G3, (G1>G3>G2)
The image recording apparatus according to claim 1, which has the following relationship. (10) The image recording apparatus according to claim 1, wherein the image type classification is such that weighting coefficients for the classification are selected and designated by operating a selection switch.