JPS6236978A - Picture processing unit - Google Patents

Abstract

PURPOSE:To reproduce a picture without losing the contrast and gradation of an original by setting a threshold value matrix in matching with the picture density of the original and its density distribution. CONSTITUTION:A density signal of the original photoelectric-transduced by an image sensor 1 is quantized through an A/D converter 2 and stored once in a memory 3. An operation circuit 4 detects a means value Dm and a deviation a of the density from the picture information read by the sensor 1 to calculate the minimum value Dmin=Dm-n.sigma of the threshold value matrix and the maximum value Dmax=Dm+n.sigma, where (n) is a preset constant. The threshold value matrix element Di=Dmin+{(Dmax-Dmin)/S]Xi is decided in matching with the weight of the threshold value, where (i) is a dot number and S is the maximum value of (i). A discriminating circuit 5 reads sequentially the picture information of the memory 3, compares the result with a threshold value calculated by the circuit 4 and gives an output to an external device such as a printer.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image processing device that expresses a halftone image using a binarized image.

(Prior Art) Devices that reproduce halftone images such as photographs using binary black and white images have come into wide use.

Various attempts have been made to reproduce halftone images with high quality as binary images.

Halftone expression using binarized images can be roughly divided into dithering methods and density pattern methods, and in the present invention, the dithering method is used. The dither method is used when the recording display resolution and the reading resolution are equal. In the recording stage (binarization processing), much of the information obtained in the reading stage must be lost. In the dither method, the read density of each pixel is compared with a threshold value calculated according to a certain rule, and then binarized. In the systematic dither method, which is excellent in gradation expression, a threshold value matrix made of pseudo-random numbers is used, and the threshold values are changed in correspondence with the coordinates of pixels. The order of the threshold values in the matrix includes Bayer type (dot dispersed type), halftone type, and spiral type (dot concentrated type).

Tables 1 to 4 show examples of threshold matrices.

Table 1 BAYER type 17 gradations (4x4) Table 2 Halftone type 17 gradations (4x4.) Table 3 Spiral type 17 gradations (4x4) Table 4 65 gradations (8x8) (To be solved by the invention Problem) In pseudo halftone reproduction, it is necessary to convert the threshold values according to each individual document and perform binarization using an optimal threshold value matrix.

Conventionally, when determining a threshold value, it has been determined by sliding a reference threshold value based on the white level, average density, density histogram, or the like of the image information of the document.

However, with these methods, it has been impossible to obtain an appropriate threshold value that does not impair the gradation of the original, such as microfilm, whose image density varies depending on the individual original.

For example, in Japanese Patent Application Laid-open No. 60-25835,
Once a document (a dummy blank sheet is also acceptable) is read, the maximum value of each bit of the image sensor is stored as the white level. From the second time onwards, the output of the image sensor is A/D converted based on the value of this white level memory, and the density gradation is quantized. Furthermore, Japanese Patent Application Laid-Open No. 60-25372 discloses an optical sensor array that detects the average brightness of light incident on the image sensor, which is arranged close to the optical sensor array for reading the original and the image sensor. , the reference threshold for binarization is changed according to the output signal of the optical sensor.

However, with these methods (the J1 threshold matrix simply slides in accordance with changes in the white level and density average value, the reproduced image does not properly follow the gradation of the original, and there is a risk of damaging the gradation. be.

Furthermore, in Japanese Patent Laid-Open No. 60-37878, the standard tone curve data is modified so as to pass through the highlight points and shadow points of the original, and when this modified data is used, the output signal corresponding to the input signal is Standard tone curve data that minimizes the sum of the deviations between this and the corresponding desired output signal is set as gradation conversion data.

However, with this method, it is necessary to have several types of standard tone curve data.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing apparatus capable of reproducing an image without impairing the gradation or contrast of an original in pseudo-halftone reproduction.

(Means for Solving the Problems) An image processing device according to the present invention includes an average value calculation unit that calculates an average value of pixel densities of input image data, and a deviation value of a density distribution of pixel densities. a deviation value calculation means; a threshold matrix setting means for setting a threshold matrix in a density range set by a predetermined method from the average value and deviation value; The present invention is characterized by comprising a binarization means for converting into values.

(Function) The threshold matrix is set according to the image density of the original and its density distribution.

(Example) Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

FIG. 2 shows the configuration of the image processing circuit. In the image reader, a density signal of a document photoelectrically converted by an image sensor (CCD) 1 is quantized through an A/D converter 2 and temporarily stored in a memory 3. The arithmetic processing circuit 4 detects an average density value and a deviation value from the image information read by the image sensor 1, and calculates a threshold value matrix. The discrimination circuit 5 sequentially reads the image information stored in the memory 3, compares it with the threshold value calculated by the arithmetic processing circuit 4, and outputs the image information from an external device such as a printer (not shown).
Output to.

Next, a method of calculating a threshold matrix performed within the arithmetic processing circuit 4 will be described. The average value D and deviation value σ of image density are calculated in advance. First, the minimum value D of the threshold matrix and Dma,1n are determined as follows.

D=D-1-n·σ.

max m Here, n is a constant that is set in advance, and a mechanism that allows the operator to freely change it when it is desired to change the gradation of the image may be used.

This threshold calculation method can be applied to any 7-trix threshold method.

Next, a threshold matrix is determined according to the threshold values.

D = D.

0 mIn.

Here, i is the number of dots (ie, a number representing the order in the threshold matrix), and S is the maximum value of i. Alternatively, there is also a method of determining the minimum value D and the increase rate ΔD of the weight, and then determining the threshold value.

D, = Di-1 + ΔD.

(However, D=D,) o man FIG. 3 shows an example of a density histogram of a document image.

FIG. 1 shows the relationship between the threshold level and image density determined by the method described above for the original image shown in FIG. Here, n was set to 2. The threshold level D is D −
l m from 2σ to D +2σ
Do to D8 for m image density
It changes linearly up to. Also, the image density is D +2σ
or D or less than D −2σ.

s Keep it.

Next, the setting values of the threshold matrix will be explained for the four types of images shown in FIGS. 4(a)-(c) and FIG. 5.

In FIG. 4(a), the image density is distributed between 0.4 and 1.6, and the average value and deviation value are 1°00 and 0.346, respectively. In FIG. 4(b), the image density is distributed between 1.0 and 1.6, and the average value and deviation value are 1.30 and 0.173, respectively.

In FIG. 4(c), the image density is distributed between 1.0 and 1.6, and the average value and deviation value are 1°31 and 0.154, respectively. In Figure 5, the image density is 0.
．． distributed between 3 and 0.4 and between 0.6 and 1.6,
The average value and deviation value are ■, lO, and 0.297, respectively. Here, the density between 0.3 and 0.4 is noise due to dust, scratches, etc. that exist in addition to the image density region (0.6 to 1.6).

For comparison, Figure 6 (a) and (b) show the image density of 0.
．． The effective gradation range is shown when the threshold matrix is set to 64 equally divided values between 4 and 1.6. When this threshold matrix is used, almost any original document can be binarized.

For example, for the image density distribution shown in FIG. 4(a), binarization is performed over the entire gradation range. However, Fig. 4 (
For documents with narrow density distributions shown in b) and (c), the threshold matrix area is too wide for the image density area, so the effective number of gradations (32) is smaller than the number of matrix gradations (64).
) and it is not possible to obtain good results.

Further, when the threshold value matrix is set between the maximum value and the minimum value of image density, it is effective for the image density distribution shown in FIGS. 4(a) to 4(c). However, if there is noise in addition to the image density region as shown in FIG. 5, the threshold value will be set up to the noise level as shown in FIG. 6(c). Therefore, not only the number of effective gradations is reduced to 50, but also noise may be output to the image.

FIGS. 7(a)-(d) show the threshold values set by the method of this embodiment for the concentration distributions shown in FIGS. 4(a)-(c) and FIG. 5. Here, n=2. The number of effective gradations is 5 in each of FIGS. 7(a) to (d).
6.56.63.54. The effective numbers of gradations according to the conventional setting method shown in FIGS. 6(a) to 6(c) corresponding to these values are 64.32.32.50, respectively. For the original density distribution shown in FIGS. 4(a) to 4(c), in this example, the number of effective gradations is always 56 or more, which is better on average than when the threshold value is fixed. results have been obtained. This is because in this embodiment, not only the threshold value matrix is slid, but also the interval between the threshold values is changed in accordance with the document, so that the gradation is not impaired. Furthermore, even for a document containing noise as shown in FIG. 5, the present embodiment has a larger number of effective gradations, and there is no risk of noise being output to the image.

In Fig. 8(a)-(c), I 1mmX 31.25
The output result to a printer after enlarging a halftone image of a microfilm of 1.0 mm in size by about 3 times and processing it by the method of this embodiment is shown. The data was read for 88,000 (176x500) dot pixels with a sampling pitch of 1/6 mm. Regarding Figures 8(a) and (b), please refer to the 4th
Using the 8x8 threshold matrix in the table, it was set as n-2. The number of data used to calculate the deviation value is approximately 4000 (5%) in Figure 8(a), and in Figure 8(b),
All data were used. Both FIGS. 8(a) and 8(b) represent halftone images with good gradation. From Figure 8(a),
It can be seen that a sufficiently good image can be obtained without calculating the deviation value from all data. In FIG. 8(C), the (4,X4) threshold matrix in Table 2 was used.

High resolution images are obtained.

On the other hand, FIG. 9(a) shows the same original as used in FIGS. 8(a)-(c) in the general density range (about 0.3-1
．． 6) is divided into 64 equal parts and the threshold values are set.

Compared to FIGS. 8(a) and 8(b), both contrast and gradation are poor. Further, FIG. 9(b) shows the result of setting threshold values by dividing the central part of the above density range into 64 equal parts.

The gradation is not as good as in FIGS. 8(a) and 8(b).

In the embodiment described above, the number of gradations of the matrix may be changed depending on the deviation value, and the number of gradations of the matrix may be changed depending on the deviation value.
An appropriate type may be used among f3r''l, halftone type, spiral type, etc. depending on the deviation value or the purpose of the output image.

A small deviation value means that the number of intermediate tones is small, so it is necessary to select a matrix with a small number of tones, or simply perform a binarization process (
It is possible to change the threshold level of the matrix to the same value. Conversely, in order to increase the resolution, B a
The yer type can also be applied. Conversely, a large deviation value means a large number of intermediate tones, so in order to improve reproducibility, a matrix with a large number of tones may be selected. On the contrary, contrast I
If you want to emphasize , you can reduce the number of gradations.

The deviation value of image density can also be used to determine whether a document has contrast or not. If the deviation value is smaller than a certain reference value, it is determined that the document has little contrast, and is binarized using a 16-gradation matrix to obtain image information with high contrast resolution. On the other hand, if the deviation value is large, it is determined that the document has contrast, and is binarized using a matrix of approximately 64 gradations to obtain image information that does not impair gradation. Thereby, image information with contrast can be obtained from a document without contrast.

(Effects of the Invention) Since the threshold value is set according to the image information and its density distribution, the image can be reproduced without impairing the contrast or gradation (smoothness of the density distribution) of the original.

[Brief explanation of drawings]

FIG. 1 is a graph showing the levels of the threshold matrix set for the concentration distribution of FIG. 3. FIG. FIG. 2 is a block diagram showing the configuration of the image processing circuit. FIG. 3 is a graph showing the distribution of image density. 4(a) to 5(c) and FIG. 5 are graphs showing the distribution of image density, respectively. FIGS. 6(a) to 6(C) are graphs showing the levels of the threshold matrix set by the conventional method. FIGS. 7(a) to 7(d) are graphs showing the levels of the threshold matrix set by the method of the present invention, respectively. FIGS. 8(a) to 8(c) are diagrams of binarized output results obtained using the threshold matrix set according to the present invention, respectively. FIGS. 9(a) and 9(b) are diagrams of binarized output results obtained using threshold matrices set by the conventional method, respectively. Patent applicant Minolta Camera Co., Ltd. Representative
Person Patent attorney: Mr. Hajime and 2 others Figure 8 (b) (C) Figure 2119 (b)

Claims (1)

[Claims] (1) An average value calculation means for calculating an average value of pixel densities of input image data; a deviation value calculation means for calculating a deviation value of a concentration distribution of pixel densities; threshold matrix setting means for setting a threshold matrix in the density range set by the method;
An image processing device comprising: binarization means for converting into values.