JPH02884A - Color image processing device - Google Patents

Abstract

PURPOSE:To almost perfectly discriminate a base part from a character part and to improve reproducibility by appropriately using density histograms for each color and a general density histogram and setting a threshold value. CONSTITUTION:A histogram creating circuit 100 creates density histograms on a color basis and the general density histogram from the former histograms. When a CPU 160 processes data on the created density histograms, it calculates a threshold (admax) indicating the base part, from the general density histogram. According to the color-by-color density histograms, an ordinary original, a reserval original, or an original with mixed density, is discriminated to set a threshold corresponding to the discriminated original. Then, a threshold that corresponds with the type of the original and the shape of the density histogram is calculated on a color basis. According to the calculated thresholds, image data is multi-valued. Thus, the base part of a color original is specified correctly.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention is a color image processing device suitable for application to a color copying machine for recording on plain paper, and in particular, a color image processing device that can set a threshold value suitable for various types of originals. The present invention relates to a color image processing bag 2.

[Background of the Invention] In a color image processing device, such as a color copying machine using a Lebbeam 2, information on a color document is separated into a plurality of colors to obtain color image information, and a color image is generated based on this color image information. I try to record it.

In addition to being able to perform various image processing such as scaling processing and partial color conversion processing, such color copying machines are also capable of converting image data that has undergone image processing into binarization or multi-value processing. Based on the binarized or multi-valued image data, an optical signal applied to the photosensitive drum as image exposure is modulated. After being formed into a latent image by this optical signal, it moves to a developing process.

An example of a color image processing apparatus that records a binarized or multivalued image after performing various types of image processing as described above is shown in FIG. 12 and subsequent figures.

Information (optical image) of a subject 2 such as a document passes through an optical system 3 and is separated into two color-separated images by a dichroic mirror 4. In this example, the image is separated into a red R color separation image and a cyan Cy color separation image. Therefore, the cutoff wavelength of the dichroic mirror 4 used is approximately 540 to 600 nm.

The color separation images of red R and cyan cy are supplied to image reading means, for example, CCDs 6 and 7, and image signals R and C, which are color image information of red component R and cyan component Cy, are respectively supplied.
y is output.

The image signals R and Cy output from each CCD 6 and 7 are A/
By being supplied to D converters 10 and 11, it is converted into a digital signal of a predetermined number of bits, 6 bits in this example. Furthermore, shielding correction can be done simultaneously with A/D conversion. 12.13 shows a shading correction circuit.

The shading-corrected digital signal is extracted by the effective area extraction circuit 15 for only (g) of the maximum original size width, and is supplied to the next stage color discrimination circuit 20. For example, if the maximum original size width to be handled is 84. f size, the gate signal is the timing signal generating means 170.
The size signal B4 generated in is used.

Here, if the shading-corrected digital drawings (g are VR and VC, respectively, then these digital signals VR and V
C is supplied to a color discrimination circuit 20 and discriminated into a plurality of color signals.

In this example, a case is illustrated in which the device is configured to discriminate into three color signals: red, blue, and black.

Color discrimination in the present invention refers to reading information from a color document and then assigning each pixel to one of, for example, one color, such as helmet, red, or black.

That is, no matter what color the original is, one color of red, helmet, or black is determined for each pixel. These red, blue, and black are generally determined by the configuration of the color image forming apparatus (for example, a color copying machine with red, blue, and black developing devices), but the types and number of colors to be attributed are may be of any other type, and does not necessarily need to correspond to the configuration of the color image forming apparatus.

Each color-discriminated color signal is composed of color code data (2 and t data) indicating its color information and density data (6 and t data) indicating its density information. The data for each of these color signals is stored, for example, in a color discrimination conversion table (map) in a ROM.

FIG. 13 shows an example of this color discrimination map.

The color signals that have undergone color discrimination are transferred to a color image processing step.

First, it is supplied to the next-stage color ghost correction means 30, and color ghosts in the main scanning direction (horizontal scanning direction) and the sub-scanning direction (drum rotation direction) are corrected.

During color discrimination, unnecessary color ghosts (especially around black characters)
This is because color ghosts) may occur.

FIG. 14 shows an example of appearance of color ghosts. The figure shows the color ghost that appears after color discrimination is performed by capturing an image of the kanji character "sexuality" written in black.

As you can see from this example, as a color ghost,
As shown in FIGS. 15A to 15C, red and blue appear at the edge of the black line, black appears at the edge of the blue line, and black appears at the edge of the red line.

For other color combinations, it is clear that color ghosts appear in five different ways.

This color ghost correction means 30 is a circuit for correcting such color ghosts as much as possible.

Color ghost correction applies only to color coat data.

In addition to color ghost correction, the image processing includes resolution f/minority correction, partial color conversion processing, scaling processing, multi-value processing, and the like.

Next, the color signal (color code data and density data) after the color ghost 11n is sent to the resolution correction circuit 40 at the subsequent stage.
At , the density data is processed and the resolution (MTF)
is corrected.

This deterioration in resolution includes deformation of the beam shape of the laser beam, deterioration of the development characteristics of toner onto the photoreceptor drum, and the like. Furthermore, it is the optical system (original reading system) and its traveling system that directly affect the deterioration of resolution.

FIG. 16 shows MTF values (before correction) in the main scanning direction and the sub-scanning direction when the optical system is driven. This characteristic is 2 to 16
These are the measured values when scanning a black and white pakuon with a spatial frequency up to dots/mm.

The MTF in this case was defined and used as MTF=(W-BK)/(W+BK)(%). Here, W is a white signal, BK is a black signal 3
This is the number.

The deterioration of MTF is more significant in the sub-scanning direction. To make the same amount of correction, the amount of correction in the sub-scanning direction should be 2 times that of the main-scanning direction.
It should be set to ~4 times.

In order to correct the main and sub-scanning directions to the same degree and not to deteriorate the reproducibility of fine line parts, the resolution correction circuit 4
0, a convolution filter using 3×3 pixel image data or the like can be used.

FIG. 17 shows the correction results when using a convolution filter.

The resolution-corrected density data, color ghost-corrected I, and color coat data are each supplied to a color data selector 50. When the partial color conversion mode is selected, that image area is recorded in a specific color.

This partial color conversion mode is a mode in which an area surrounded by a marker (color marker) on a black and white document is recorded in the color of the color marker.

For example, as shown in FIG. 18, in this mode, an area a surrounded by a blue color marker is recorded in blue.

To do this, it is necessary to detect the color marker and extract its area.

For this reason, the area extraction circuit 60 is provided.
The area of the color marker on the paper is detected, and the area signal is supplied to the color data selector 50.

The region extraction circuit 60 outputs region signals QR and QB corresponding to the color marker region, as shown in FIG. 19, for example.

In addition to these signals, the color data selector 50 also receives
A scan code signal indicating which color is currently being copied and a partial color conversion command signal CC are respectively supplied.

This is a multi-color copying machine that can record multiple specific colors.One color is developed for each rotation of the photoreceptor drum, and after all colors have been developed, a transfer separation process is performed. In the case of a type that records a color image by doing this, the scan code signal indicates which color is currently being developed.

Therefore, when a blue color marker is detected, if the corresponding color image is output during the blue copy sequence and when the area signal is obtained, the image within the blue color marker can be output. It can be recorded in blue.

When it is not a partial color conversion process, density data is output only when the color code data matches the scan code (g).In other words, in the case of a red copy sequence, while the red color code is obtained, The corresponding density data is selectively output.

The image data output from the color data selector 50 (
density data) is input to the variable magnification circuit 70, where it is enlarged and
The reduction process is it! ! be done.

Enlargement/reduction processing is performed by interpolating the density data in the main scanning direction, and by controlling the scanning speed in the sub-scanning direction (rotation direction of the photosensitive drum).

If the scanning speed is increased, the sampling data in the sub-scanning direction is thinned out, resulting in a reduction process; on the other hand, if the scanning speed is decreased, the data is enlarged.

The density data subjected to the enlargement/reduction processing is sent to the multi-value conversion circuit 80.
The data is input to , and multi-value processing is performed. For example, by using four thresholds, the density data of a 6-pit configuration can be
It can be valued.

Threshold data is set manually or automatically.

A histogram creation circuit 100 is provided to automatically determine threshold data.

In the histogram creation circuit 100, a density histogram of the entire document is created from the total density information of the document, and optimal threshold data for the image is calculated based on the created density pistogram.

The 3-bit multi-value data subjected to multi-value processing by the multi-value processing circuit 80 is supplied to the driver 140 via the interface circuit 130.

The driver 140 can modulate the laser beam in accordance with the multilevel data. In this example, PWM modulation is possible.

The driver 140 may be built into the multi-value conversion circuit 80.

Output device 150 by PWM modulated laser beam
A latent image is formed by irradiating the photoreceptor drum provided with the double exposure light, and the latent image is developed by operating the developer (12g) containing toner of a predetermined color.

As the output device 150, such a laser recording device or the like can be used. In this example, two-component non-contact jumping development and reversal development are employed.

That is, the transfer drum used in conventional color image formation is not used. In order to reduce the size of the device, three-color images of blue, red, and black are developed on an OPC photoreceptor (drum) for image formation in three rotations of a tram, and transferred once after development to print on plain paper. I try to transfer it to recording paper such as.

The various image processing commands and image processing timings described above can all be controlled by the CPU 160.

Reference numeral 170 denotes a processing timing signal generation circuit for obtaining various processing timings, which includes a clock CLK as well as horizontal and vertical synchronizing signals H, H, and vertical synchronizing signals in the main scanning direction and sub-scanning direction obtained from the output device 150 side. V,
An index signal IDX indicating the start of laser beam scanning can be supplied to V-V Kira.

180 is a timing signal generation circuit for obtaining scaling timing.

[Problem to be solved by the invention] In the above-described configuration, the image data is binarized or multi-valued by the multi-value conversion circuit 80 and then supplied to the output device 150 side. . It is preferable to record an image based on multivalued multivalued data because halftones can be reproduced well.

When image data is multivalued in this manner, a threshold value suitable for the image can be set based on the density histogram data created by the histogram creation circuit 100 as described above.

For example, if the output device 150 is a monochrome blinker, the density histogram as shown in FIG.
I am trying to set two thresholds. Based on this set threshold value, the background portion and character portion of the document are separated.

When a color printer is used as the output device 150, it is preferable to perform multi-value conversion based on threshold values set for each color.

This is because, as shown in FIG. 21E, it may not be possible to clearly distinguish between the background portion and the character portion using only the overall density histogram based on the total density information of the document. This C is particularly noticeable when it is desired to set a plurality of threshold values. In such a case, if a density histogram is created for each color, the density histogram will look like the one shown in FIG.

This is because the density histogram for each color often has a relatively simple shape.

In addition, in FIG. 21, the black and white density histogram is divided at a brightness level of, for example, "41".
The following is a black density histogram (see the same figure), and "41"
The above is shown as a white density histogram (D in the same figure).

Furthermore, as shown in FIG. 22, the brightness level may vary for each color, and in such a case, unless a threshold value is determined for each color, the reproducibility of the image will deteriorate significantly.

The reason why the brightness level varies for each color is as follows.

The sensitivity characteristics of a CCD, which is a photoelectric conversion element provided in a document reading system (optical system), do not match human vision. Therefore, even if the color originals have the same density, that is, the colors appear the same to the human eye, when a color document is read by an optical system, the brightness level will not be the same. the result,
This is because the density histogram also has different characteristics for each color.

The example shown in FIG. 22 is an example in which a density histogram is created for each color by imaging the text portion of an English chart with the same density of black, blue, and red.

Therefore, in order to equally reproduce each color, it is necessary to create a density pistogram for each color and determine the threshold for each color, rather than determining the threshold from the overall density histogram.

Considering this, it may be possible to determine the threshold value for each color from the density histogram for each color in all cases.

However, if the background part of the manuscript is not composed of a single color, such as old newspaper, the density histogram for each color will be as shown in Figure 23, and the density histogram for each color will be There may be cases where it is not possible to identify the background part.

Further, since normal originals have black characters written on a white background, there are many originals whose text density is darker than the background density (hereinafter referred to as normal originals) (see FIG. 24A). However, there are some original f heights (hereinafter referred to as inverted originals) that are lighter than the character density or background density, such as white characters (Figure B
When determining the threshold value, it is necessary to appropriately respond to both types of manuscripts.

This is because even in the case of a reversed original, if it is processed in the same way as in the case of a normal original, a blank copy will result.

Roughly, depending on the height, characters with a clear 0-reverse difference, such as originals with mixed shading (see Figure 25 A) where the density of the characters themselves are uneven, can be either dark or light depending on the threshold setting. Area X) may be blurred.

Therefore, in this invention, taking these points into consideration, calculations and judgments are performed based on the data of the density pistogram for each color and the overall density histogram, and threshold values suitable for various types of originals can be set. This paper proposes a color image processing device.

An object of the present invention is to accurately specify the background portion of a color document and to set a threshold value for each color.

Another object of the present invention is to prevent blank copies by determining whether a color original is a normal original or a reversed original.

Still another object ζ of the present invention is to determine whether a color document is a mixed-shade document and to prevent misalignment of low-density characters, etc.

[Means for Solving the Problems] An object of the present invention is to provide a color image processing device that performs image processing on color image information based on color image information converted into an electrical signal. A color discrimination means for image information and a threshold setting means are provided, and the threshold setting means creates a density histogram for each color, and also creates a total density pistogram from the density histogram for each color. The density corresponding to the background part of the document is detected, and a threshold value for color image information is set for each color from the 53 degree histogram for each color based on the density data of the background part. Roughly speaking, it is possible to determine whether it is a normal original or a reversed original, perform the corresponding processing, and set a threshold value.

Roughly, from the density histogram of each color, PK light R Arihara I
The best feature is that a different threshold value is set when a light and dark document is detected.

[Operation] In this configuration, the pistogram creation circuit 100 creates a density histogram for each color, and also creates an overall density histogram from this density histogram.

The 1-overall density histogram is a histogram obtained by adding the frequency data of the 1-degree histograms for each color.

By processing the created density histogram data by the CPU 160, a threshold value (adma
It will be done.

Therefore, the threshold setting means is composed of the histogram creation circuit 100 and the CI 160.

Furthermore, a standard document, a reversed document, or a mixed document of light and shade is determined from the color density histogram, and a threshold value is set accordingly.

When setting the threshold value, it is determined whether the shape is a density pistogram of six colors.

Based on various experiments and results, it is preferable that the shape of the histogram is classified into three types: background type, character type (mixed type), and other types that do not belong to these types, depending on the content of the color manuscript.

Next, a threshold value is calculated in accordance with the type of document and the shape of this density histogram. To exemplify the case of converting image data into five values, in this case, multi-value conversion is performed using four threshold values, so it is necessary to calculate four threshold values (threshold value 1 to threshold value 4).

These threshold values are calculated for each color, and the image is multi-valued based on the calculated threshold values.

Note that if the shape of the concentration histogram is recognized as a shape other than the pond shape, the threshold value can be selected from among the multiple threshold values used in manual settings, as in the case of other shapes. It is important that the equipment has such functions as possible.

[Example] Next, a case in which the present invention is applied to the above-mentioned color image processing device will be described in detail with reference to FIG. 1 and subsequent figures.

FIG. 1 shows a specific example of a color image processing apparatus 1 according to the present invention, and the same parts as in FIG. 12 are given the same reference numerals, and their explanation will be omitted.

At opening of the park, the color code data corrected for color ghosts or the density data are supplied to the scaling circuit 70, and the scaled density data and color coat data are supplied to the multi-value conversion circuit 80, respectively. Whether or not to perform color ghost correction processing is controlled by the CPU 160. When there are multiple color discrimination tables, the CPU 160 selects which table to use.

In the present invention, the creation and selection of threshold data for multi-value processing is performed by the CPU 160. Therefore, the threshold value setting means is comprised of the histogram creation circuit 100 and the CPU 160 in terms of hardware.

The histogram creation circuit 100 creates a density histogram for each color and calculates threshold data for each color in order to perform optimal multi-value processing for each color. Therefore, color code data is also supplied to this histogram creation circuit 100 in addition to density data.

In this example, when creating the density histogram, the second
As shown by diagonal lines in the figure, image data is obtained by performing Briscan PS on a predetermined image area (a part of the entire image) of the document.

The reason why density data is detected by performing Briscan PS on only a predetermined image area is that the density data necessary for creating a density histogram can be collected from a predetermined image area without requiring a significant amount of Briscan time. This is for the purpose of

In this embodiment, the briscan time is set within the copying machine's copying-up time. The reason why warming-up is required for copying (k) is as follows.

First, the light 11 irradiated onto the photoreceptor drum for copying (etc.)
Rotating polygonal surface (polygon mirror, etc.) for deflecting and scanning No. 9 (a signal modulated by a multi-inl signal, such as a sea 1 beam) in the main scanning direction.
Or is provided. In order to stabilize the rotation of the rotating polygon when the power is turned on, some warm-up time is required.

Second, it is necessary to pre-rotate the photosensitive drum (for example, within one rotation) to stabilize the copying process.

Since these warm-up times are about 2 to 3 seconds, it is preferable that the briscan be set to end at a time close to this warm-up time, preferably within the warm-up time.

However, if such a time is set as the blisscan time, the area of the predetermined region to be blisscanned becomes extremely narrow.

To avoid this drawback, in the main scanning direction (horizontal direction in Figure 2)
The document information necessary for threshold calculation is extracted by sampling the document information while thinning it out, and by making the scanning speed of the optical system faster in the sub-scanning direction (vertical direction) than during normal recording.

For example, when the resolution is 16 dots/mm, 1
Information on the document is sampled in the horizontal direction while being thinned out to about dots/mm.

In this example, the scanning speed in the sub-scanning direction was selected to be approximately 4 times the speed. By setting the speed to 4x, the information of the document is sampled once every four lines. Furthermore, the averaged image data of the document information within a predetermined area by quadruple speed scanning is detected as density data.

When selected in this way, assuming that the maximum original size is B4 size, in the embodiment, as shown in FIG. The range is .

Image data detected within this range is used to create a density histogram.

In order to create a density histogram for each color, the memory (not shown) provided in the histogram creation circuit 100 stores frequency data in memory areas divided for each color, as shown in FIG. Stored.

For this reason, color code data for classifying memory areas is also used as upper address data for the memory (see FIGS. 3 and 4).

After the density histogram for each color is created, an overall density histogram is created using the density histogram data for each color, and thresholds 1 to 4 are calculated by the CPU 160.

An example of threshold value calculation will be explained with reference to FIG. 5 and subsequent figures.

First, a density histogram for each color is shown in FIG. 5 for convenience. FIG. 6 shows an overall density histogram created from the density histograms for each color.

FIG. 7 shows an example of a flowchart for calculating a threshold value from a density histogram.

First, an overall density histogram is created from the density histograms for each color created by the histogram creation circuit 100 (step 300).

Next, data at the point [· shown in FIG. 6] is calculated from this total 4 degree histogram (step 310).

The necessary data is the maximum degree admax and the maximum brightness level aw! Qax and the frequency at the foot of the mountain 2 (
The brightness level at that time (the base brightness level) is awmin.

Then, from each frequency data of the density histogram for each color,
Similarly, necessary data for the necessary 8=degree histogram is calculated (step 200). This processing also includes processing for originals with mixed shades of light and light.

The data required for the density histogram include the maximum and minimum values wh, w, the maximum brightness level Wmax between the maximum frequency dmax and t, and the minimum frequency dmi, as shown in Figure S.
n and the brightness level W min at that time.

After calculating the necessary data for the density pistogram, the shape of the histogram is determined for each color (step 21).
0).

In this shape determination step 210, a normal document and a reversed document are discriminated, and these are separated into a background type, a character type, and other types not belonging to these types.

Here, in a normal document, that is, a normal document, the background portion has substantially uniform density and is larger in area than the character portion. Also,
Since the density is brighter than that of the text portion, the histogram has an elongated mountain shape, as shown in Figure 21, and is on the brighter side overall.

Therefore, if a color-specific 0z degree histogram exists at a level on the brighter side than the mountain foot portion detected in the overall δ1 degree histogram, the data of this density pistogram is considered to be the background.

Therefore, it can be set as the frequency data of the mountain foot portion Z (the mountain foot frequency) or the threshold value of the background portion.

On the contrary, the text portion has a more uneven density than the background and occupies a smaller area, resulting in a flat, mountain-shaped histogram. Therefore, when the frequency data of the density histogram for each color exists at a level on the darker A side than the foot part Z (Fig. 21 A, B), or when the frequency data exists across both sides (Fig. 23 A) ), the side darker than the set threshold of the background portion is regarded as a character, and a threshold is set for this area.

Then, histograms with shapes that do not belong to these shape examples were classified as other types.

Inverted pieces such as white characters (higher ones do not have their own toner, so
The background part must be output as black, for example. 7
This is because if the background portion is removed as in the case of a single original, it will become a blank copy.

Incidentally, since the density relationship between the background and characters in a reversed manuscript is reversed from that of a normal manuscript, as shown in Figure 24B, the original (:
The background portion of the text (°) exists on the darker side (left side) of the character portion in the pistogram L.

Therefore, in this process, if it is determined that the document is reversed,
A threshold value was set for the background part as a character type.

When the shape determination of the density histogram is completed, the optimal No. 1 value is extracted for each shape (step 230).

Since these steps 200 to 230 are sequentially processed for each color, when the processing is completed for all colors (step 240), the threshold value determination processing routine ends.

FIG. 8 shows a processing routine for calculating each point of the total density histogram.

In the same figure, the maximum degree ad ma
The maximum brightness level aWmax between x and then .times. is calculated (step 311).

Next, the foot of the mountain Z can be detected from the maximum frequency admax side (step 312). The detection of the foot of the mountain Z is performed by comparing the frequency data of the previous i(), and when the latest frequency data is larger than the frequency data of the next two times in a row, the frequency data two degrees before that is determined as the foot of the mountain Z. is used as frequency data admax.

The brightness level aWmin at this time is used as a boundary level for dividing the text portion and the background portion on the histogram.

FIG. 9 shows a processing routine for calculating each point of the density histogram for each color.

In the same figure, in this example, the minimum value W, 11 and the maximum value wh
is calculated, and in order to handle manuscripts with mixed shading, Hiss!
・Detection is performed for the hollow frequency in grams, and when a hollow frequency is detected, the minimum value W dan is corrected to W fl a as shown in FIG. 25B (step 201.2
02,250).

Below, the maximum degree dmax and the maximum brightness level W at that time
maxX minimum power dmin and the minimum luminance level W min at that time are calculated in this order (steps 203 and 204).

Here, the minimum value Wl and maximum value W h of the color-based histogram
When detecting fl! at the following concentration level: The threshold value setting range was determined within the following range.

Black: Blue within 4 to 4 levels Red within 15 to 4 levels Red within 25 to 45 levels This is in line with the density characteristics of the actual device, and is intended to eliminate waste in calculation processing and optimize the threshold value. be. In this example, a level corresponding to a practical a degree of 1, 0 to 0°1 was determined for each color as described above, and the results were similar.

Below, we will explain an example in which a mixed-density original is determined, and after determining whether it is a full-length original or a reversed original, it is determined whether it is a background type, character type, or σQ other type, and the threshold value is set accordingly. do.

FIG. 10 shows a processing routine for determining whether an original is a mixture of light and shade and changing the minimum value W to Wlla if the original is a mixture of light and shade. Whether or not the original is a mixture of shading and lightness can be determined by detecting hollows in the frequency.

First, in order to determine whether the character histogram includes a portion of light density, the level position of the maximum value wh of the histogram is determined. The standard concentration that is the basis for judgment is O~3
Set within the 0 level. In this example, it is set to 30 (step 251). The following processing is not performed for sharpeners with a density higher than the set value of 4a degrees.

If the pistogram contains a light density, the count value i of the counter is set to the reference density value (=30) (step 252) in order to detect whether there is a hollow part or not (step 252), and the frequency at the count value i is It is checked whether or not is greater than or equal to the foot cut frequency (step 253).

The cutting frequency and the frequency for removing noise,
If the fluctuation range due to noise is considered to be within 0.1% (approximately 64 in frequency value) of the total number of pixels (approximately f34.ooo in this example) existing in the Briscan area shown in Fig. 2. ,0.
This cutoff frequency may be set within 1%. In this example, it is set to 50.

If the height is less than or equal to the cutting frequency, the difference between the count value i and the minimum value WJl is compared, and it is determined whether the difference is large or small (step 256). If the difference between the two is larger than a predetermined value, it is assumed that a hollow part exists, and the minimum value is changed.

Here, the predetermined value is a value that takes multivalue processing into consideration,
When performing 5-value conversion, a difference of at least 5 levels is required. In this example, the predetermined value is set to 5. When the predetermined value is 5 or more, the minimum value W is changed to the count value i (step 257). That is, at this time, the force 1 kundt value i is used as the corrected minimum value W (=WJ1a).

If the height is less than a predetermined value, the calculated minimum value W is used as is, since there is no effect of increasing the height even if the process is performed.

When the frequency at count value i is greater than or equal to the cut-off frequency,
In order to move the frequency toward a higher density side, the count value i can be decremented (step 254). Then, the magnitude relationship between the count value i at that point and the minimum value W is checked, and when the count value i is larger than the minimum value W,
Returning again to step 253, the same processing as described above is performed.

When the frequency at the count value i becomes equal to or less than the cut-off frequency in this processing step, the process moves to step 256. Similarly, in step 255, when the count value i becomes equal to or larger than the minimum value WfL, the process also transitions to step 256.

In these cases, the minimum value W1 is determined based on the count value i when the last knee F"y is performed.
1 f'7 correct processing is performed (step 256.25
7).

When a blank is detected as described above, the minimum value W is changed to W is a, and one of the thresholds 1 to 4 is determined based on this (FIG. 25B). As a result, there is no fear that particularly light text will fly out.

Figure 11 shows density histogram shape determination processing routine 2.
An example of 10 is shown below.

Shape determination is performed based on data of each point calculated in processing routines 200 and 300.

First, it is determined whether the maximum degree dmax is reached (step 211
).

In step 211, it is determined whether the shape of the density histogram for each color belongs to either a background type or a character type. It is determined whether frequency data of the density histogram exists or not. Therefore, (cutting frequency + a certain constant) (for example, 100) is used as the judgment value.

When it is determined that the background type is not a character type, and accordingly, there is no frequency data as shown in FIG. 23C, it is processed as another type (step 216).

When this type is selected, the median value used in the manual threshold value is used as the threshold value (FIG. 23C).

In this way, when the shape of the density histogram is neither background type nor character type, the reason why the manual threshold is set to the median value is to prevent the threshold from being set to an extreme or unstable threshold. It is.

Manual threshold values are also available for each color.

Next, a judgment is made as to whether the original is a dubious original or a reversed original (step 2).
13).

This judgment is based on the histograms of both manuscripts (Fig. 24A, B).
), this determination is made depending on whether the maximum luminance level W max in the background area is on the darker side or on the lighter side than a certain determination level n.

As a result of investigating the density of various types of paper used for ordinary manuscripts, it was found that this judgment level n was good at a density level of about 27 (in terms of reflection density, about 0.35).

When the maximum brightness level Wmax of the color-based histogram is on the lighter side than the determination level n, a normal document is processed.

On the other hand, when it is on the darker side than the determination level n, the inverted original is processed.

If it is determined that the original is a reversed original, the density Nakabatake (histogram 'I'm) W a is calculated (step 214
).

The histogram width Wa is Wa=Wh-W standing It is.

If the pistogram width Wa is equal to or greater than a predetermined value, it is determined that the pistogram is of a character type, and a corresponding threshold setting process is executed (steps 215, 217).

If it is determined that the value is less than or equal to the predetermined value, it is processed as another type (steps 215 and 216).

Here, the predetermined value serving as the determination criterion is the minimum histogram required to set the threshold for multi-value conversion (in the example, 5-value conversion)! - Gram convergence, and according to experiments, about 10 is good.

In this manner, in the case of a reversed document, a threshold value is set as a character type even if the background has a large frequency data in the histogram. However, if the histogram width is narrow (a small amount of histogram data), it will be processed as another type.

On the other hand, if the document is determined to be a normal document, the minimum brightness aWmi detected in the overall density histogram
n (the boundary level between the characters and the background) is compared with the minimum value W and the maximum value wh of the color-based histogram (step 218
).

Then, when the color-specific histogram is on the darker side (left side) than the boundary level aWmin, that is, when aWmin>Wh, the histogram width Wa Wa = W h - W 1 is calculated (step 219), and the character type or other treated as a type. In other words, when the histogram width is greater than or equal to a predetermined value, it is processed as a character type, and when it is less than or equal to a predetermined value, it is processed as another type (
Steps 215-217).

The predetermined value here is 1 as in the case of the reversed original.
It was good, about 0 reno.

When there is a boundary level aWmin in the color histogram, that is, W +5 ≦ aWmin ≦ wh Histogram width WaWa excluding throat and background parts =
a Wm1n -W 11 is calculated and set as a character type, and threshold value setting processing corresponding to the character type is performed (steps 220, 217).

On the other hand, when the color-based histogram is on the lighter side (right side) than the boundary level aWmin, that is, when a Wm1n<W l + 5, the minimum value WJI of the color-based histogram is replaced with the boundary level aWmin, and the entire histogram is collected as the background. (steps 221 and 222).

Through the above processing routine, a normal original and a reversed original are determined, and based on the density histogram for each color, it is determined that the shape belongs to the density histogram shape for each color or a preset shape. It will happen.

In the next processing routine 230, a threshold value is calculated and set in accordance with the shape.

The relationship between the shape of the histogram, that is, the background type, character type, and other types, and the set threshold values thereof will be illustrated below.

[Background type] In this case, the entire density histogram is regarded as background data, and a threshold value is set outside the density histogram.

Therefore, as shown in 23l8, a threshold value is set based on the minimum brightness level aWmin calculated from the total density histogram. The setting threshold value is as shown in the following example.

1 turn (direct 1 = aWmin 0 aWminX (-0,
45) Threshold 2=aWmin+aWminX (-0,3
5) Threshold 3=aWmin+aWminX (-0,20
) Threshold 4=aWmin+aWminX (-0,05)
Since the correction coefficient is negative, thresholds 1 to 4 should all be set to the left and right of the minimum brightness level aWmin (FIG. 23B).

[Character type] In this case, the density histogram range on the darker side than the base part Z of the overall density histogram is regarded as the character part.

As a result, as shown in FIG. 23A, a plurality of threshold values are set between the minimum value W and the minimum brightness level aWmin. An example of the set threshold value is shown below.

Threshold 1 = Wl + Wax (0, 05) Threshold 2 = Wf
l +WaX (0,15) Threshold 3=Wu +WaX
(0,35) Threshold 4=WJl +WaX (0,60)
These threshold values are calculated for each color, and the image data is multi-valued based on the calculated 1°C value.

In the case of a mixed-shade original, WJla may be used instead of W-tachi.

The numerical values in parentheses above are parameters for setting threshold values 1 to 4, and these values were good.

[Other types] If the shape is neither a character type nor a background type, it is processed as another type.

If any other type is selected, a predetermined or fixed value is used as the threshold (Figure 23C). Here, the median value used in manual setting is set as a fixed value. Of course, this value is just an example.

Note that the shape of the histogram is 8. If you're smart enough, you can manually set the threshold to select one of several available thresholds, just as you would for any other type.

Further, in the above example, a case where image data is converted into five values has been described, but the number of multi-value conversion may be 2 or more and is not limited to that number.

[Effects of the Invention] As described above, in the present invention, the threshold value is set by using the density histogram for each color and the overall density histogram.

By using each of the overall density histograms as well as the density histograms for each color, it becomes possible to almost accurately distinguish between the background portion and the character portion.

Therefore, it has the characteristic of greatly improving reproducibility.

Roughly speaking, since the process is performed to discriminate not only normal originals but also inverted originals and mixed-shade originals, it is possible to set a threshold value suitable for all kinds of originals.

Therefore, the present invention is extremely suitable for application to the color image processing apparatus as described above.

[Brief explanation of the drawing]

FIG. 1 is a system diagram showing an example of an image quality processing device according to the present invention, FIG. 2 is a diagram showing a Briscan area, and FIG. 3 is a diagram showing the relationship between image data and memory area V. Figure 3 shows the relationship between colors and color coat data, Figure 5 is a characteristic diagram of the density histogram for each color, Figure 6 is a characteristic diagram of the overall density histogram, and Figure 7 is a diagram of the threshold calculation processing routine. A flowchart showing an example, FIG. 8 is a flowchart showing a processing routine for calculating data at each point of an overall density histogram, and FIG. 9 is a flowchart showing a processing routine for calculating data at each point in a color-specific histogram. Flowchart 1 for skipping the processing routine, FIG. 10 shows a flowchart showing an example of the detection routine, with the outline of each color histogram shown, FIG. 11 is a flowchart showing an example of the processing routine for determining the shape of the histogram, and FIG. 13 is an explanatory diagram of a color discrimination map, FIGS. 14 and 15 are explanatory diagrams of color ghosts, and FIGS. 16 and 17 are explanatory diagrams of resolution correction. ,
18 and 19 are explanatory diagrams of partial color conversion, FIG. 20 is a characteristic diagram of density histograms, FIGS. 21 to 23 are diagrams of density histograms for each color and their total density pistograms, and FIG. The figures are f-degree histograms of a normal original and a reversed original, and FIG. 25 is a density histogram of a mixed original. 1...Color image processing device 20...Color discrimination circuit 30...Color ghost correction means, resolution correction circuit, color data selector, area extraction circuit, scaling circuit, multi-value conversion circuit, histodaram creation circuit, Interface circuit/driver/output device/CPtJ/processing timing signal generation circuit 257mm
-, I-7 74Qrt'Nr/L-4/D ＠1 capital number Q6 (QR') Monk p Figure #Monk × The ω Figure

Claims (3)

[Claims] (1) A color image processing device configured to image-process color image information based on color image information converted into an electrical signal, further comprising color discrimination means and threshold setting means for the color image information; In the threshold value setting means, a density histogram is created for each color, and a total density histogram is created from the density histograms for each color.The density corresponding to the background part of the document is detected from this total density histogram. A color image processing device characterized in that a threshold value for the color image information is set for each color from the density histogram for each color based on density data of a background portion. 2. The color image processing apparatus according to claim 1, wherein a normal original and a reversed original are discriminated from the density histogram for each color. (3) The color image processing apparatus according to claim 1, wherein a mixed density original is detected from the density histogram for each color, and the threshold value is changed when a mixed density original is detected.