JP2002171406A - Image processing apparatus, image processing method, and image processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To overcome the problem of a conventional image processing apparatus such that the image quality of image data is deteriorated because attribute discrimination is applied to each object block on the basis of orthogonal data resulting from applying orthogonal transform to the image data, resulting in increasing the adverse effect when the attribute is mis-discriminated. SOLUTION: A reference selection means 1 of the image processing apparatus of this invention selects a reference block 2R and a target pixel. An orthogonal transform means 2 applies orthogonal transform to pixel data in the reference block 2R to obtain the orthogonal data. A feature calculation means 3 calculates feature values Fm, Fn, Fp representing features of the reference block 2R on the basis of the orthogonal data. Then an attribute discrimination means 4 discriminates the attribute of the target pixel on the basis of the feature values. Thus, the image processing apparatus improves the accuracy of attribute discrimination and enhances the image quality.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, an image processing method, and an image processing program for determining attributes such as a halftone area, a character area, and a continuous tone area appearing in image data.

[0002]

2. Description of the Related Art In recent years, digital copying machines have become widespread and are replacing analog copying machines. The digital copying machine is a charge coupled device (CCD).
ce) The image of the document is read by a sensor or the like, and the signal of this image is converted into digital data. Further, the digital copier performs image processing such as filtering, gamma processing, and gradation processing on the obtained image data,
The image data is recorded on paper by a laser beam printer or the like.

Various areas such as a halftone area, a character / line drawing area, and a continuous tone area appear in an image of a document. Image processing for recording an image of a document on a sheet with high image quality differs depending on which region an image or part of an image belongs to. For this reason, some digital copiers determine the attribute to which area the image data belongs, separate the data according to the attribute, and vary the image processing for each separated data. .

Various methods for performing such separation based on image data expressed in a spatial domain have been conventionally devised. Even when data expressed in the spatial frequency domain by orthogonal transform such as FFT (Fast Fourier Transform) or DCT (Discrete Cosine Transform), an inverse transform is sometimes performed once to apply a conventional method. . In this case, after the separation is performed based on the image data expressed in the spatial domain by the inverse transform, the orthogonal transform is performed.

On the other hand, there is also a method of performing the above-described separation based on data expressed in a spatial frequency domain. For example, JP-A-6-223172 describes a technique for separating a character region and a non-character region based on data obtained by performing wavelet transform on image data. Also, Japanese Patent Application Laid-Open
Japanese Patent Application Laid-Open No. 1-220622 describes a technique in which a determination is made for each target block based on data obtained by performing a wavelet transform on image data, and each target block is separated into a character area, a halftone area, a continuous tone area, and the like. Have been.

[0006]

However, when a method of performing separation based on image data represented in the spatial domain is employed, when data represented in the spatial frequency domain is handled, an inverse transform and a transform for separation are performed. It is necessary to perform orthogonal transformation, which requires extra processing and time.

[0007] Further, even when a method of performing separation based on data expressed in the spatial frequency domain is adopted, an image is divided into a character region and a non-character region as in the technique described in Japanese Patent Laid-Open No. Hei 6-223172. There is a case where the image quality becomes insufficient only by the separation. Further, JP-A-11-2206
Even if the data subjected to the wavelet transform is separated into a character area, a halftone area, a continuous tone area, and the like as in the technique described in Japanese Patent Publication No. 22, the adverse effect of an erroneous determination is not obtained because the unit is the target block. And the image quality deteriorates.

As described above, in an image processing apparatus such as a digital copying machine or a scanner, although it is necessary to determine the attribute of image data in detail and perform processing suitable for the attribute, a sufficient attribute is required. The judgment has not been made.

The present invention has been made in view of the above-mentioned problems in the conventional technology, and provides an image processing apparatus, an image processing method, and an image processing program capable of performing accurate attribute determination. It is intended to do so.

[0010]

The present invention employs the following means to achieve the above object.

In the present invention, first, the reference selecting means selects a reference block including a plurality of pixels constituting an image, and selects a region of interest consisting of one or more pixels in the reference block. Next, the orthogonal transform means obtains orthogonal data as a result of orthogonally transforming the values of the pixels in the reference block. Further, the feature calculation means calculates a feature value representing a feature of each of the spatial frequency bands of the reference block based on the orthogonal data. Then, the attribute determining means determines the attribute of the attention area based on the characteristic value.

The attribute to be determined is, for example, a halftone dot area, a character area, or a continuous tone area. Since these attributes are determined for each attention area, accurate attribute determination can be performed.

[0013] Further, an integrating means for integrating the characteristic value of the reference block and the characteristic value of another block for each spatial frequency may be provided. In this case, the attribute determining means determines the attribute of the attention area based on the integration result of the integrating means. The accuracy of attribute determination can be improved by an amount corresponding to the integration. Even if the judgment result for other blocks is used to make the judgment, the effect of the erroneous judgment is likely to propagate,
Since the characteristic values of the other blocks are integrated and the attribute determination is performed only for the attention area, the adverse effect of the erroneous determination can be suppressed.

Further, among the orthogonal data, a correction calculating means for calculating a correction value based on orthogonal data in a predetermined spatial frequency band may be provided. In this case, the attribute determining means determines the attribute of the attention area based on the characteristic value and the correction value. By using the correction value for attribute determination, erroneous determination when the characteristic values are similar can be suppressed.

Further, not only the characteristic value but also the correction value may be integrated by the integrating means. Further, it is preferable that the feature calculation unit calculates the feature value except for the orthogonal data having the lowest spatial frequency band.

The image to be subjected to attribute determination is, for example, an image read from a document. In this case, the reference selecting means sequentially selects the reference blocks sequentially in the order from the main scanning direction of the document reading to the sub-scanning direction and shifted by one pixel.

Further, when pixels constituting an image are included only in a partial area in the reference block, the orthogonal transform means
The value of a pixel in an area other than the partial area in the reference block is filled with a value representing a white image.

Further, when the region corresponding to the region of interest of another block is outside the image, the integrating means does not integrate the characteristic value of the block having the region corresponding to the region of interest outside the image.
Only the feature value of the reference block and the feature value of another block having a region of interest in the image are integrated. Another block is, for example, a block whose attention area is around the attention area of the reference block.

The determination result is used for selecting image processing. The image processing includes simple binarization processing, error diffusion processing, dither processing, compression processing, and the like.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment)
For example, it appears in digital copiers. FIG. 2 shows the overall configuration of this digital copying machine.

As shown in FIG. 2, this digital copying machine has an image scanner section 1 for reading an image of a document 104.
01 is provided. A platen glass (hereinafter, referred to as a platen) 103 is provided on the upper surface of the housing of the image scanner unit 101.
Is provided. The original 104 is the platen 103
Placed on top. The lamp 105 exposes the document 104 through the platen 103. Scanning mirrors 106, 10
Reference numerals 7 and 108 guide the reflected light from the exposed document 104 to the imaging lens 109. The imaging lens 109 is used to
The reflected light from 4 is imaged on the line sensor 110.
Lamp 105 and scanning mirrors 106, 107, 108
Moves at a constant scanning speed in the sub-scanning direction Y. The scanning speed of the scanning mirrors 107 and 108 is half of the scanning speed V of the lamp 105 and the scanning mirror 106, and the optical path length does not change by scanning. The line sensor 110 is configured by, for example, a CCD or the like, and converts reflected light into an analog electric signal. Since the electrical scanning direction of the line sensor 110 is parallel to the main scanning direction perpendicular to the sub-scanning direction Y, the lamp 1
By moving the scanning mirror 05 and the scanning mirrors 106, 107, and 108, an analog electric signal is obtained over the entire surface of the document 104. This analog signal is output to the line sensor 11
0 is output to the signal processing unit 111. Signal processing unit 11
The printer unit 1 converts an analog signal into digital data, performs image processing on the obtained image data, and
02 is output.

The printer unit 102 records the image data input from the image scanner unit 101 on paper.
It is the laser driver 112 that sends the image data. The laser driver 112 modulates and drives the semiconductor laser 113 according to image data. The polygon mirror 114 and the fθ lens 115 apply laser light from the semiconductor laser 113 to the photosensitive drum 1 via the mirror 116.
Scan over 17. As a result, an electrostatic latent image is formed on the photosensitive drum 117. On the peripheral surface of the photosensitive drum 117, a developing device 118 is arranged. The developing device 118
The toner (developer) is attached to the electrostatic latent image to develop the electrostatic latent image. The transfer drum 119 includes a paper cassette 120,
The developed image (toner image) is transferred to the paper fed from 21. The sheet on which the toner image has been transferred is transported from the transfer drum 119 to the fixing unit 122. The fixing unit 122 fixes the toner image on the sheet. After fixing, the paper is discharged to the outside.

The quality of the image recorded on the paper in this way varies in the image processing in the signal processing unit 111. Here, FIG. 1 shows a more detailed configuration of the signal processing unit 111.

As shown in FIG. 1, the signal processing unit 111
An A / D converter 11 is provided. The A / D converter 11
An analog electric signal (image signal s00) input from the line sensor 110 is converted into digital image data a00. The analog electric signal corresponds to the luminance of the reflected light incident on each pixel of the line sensor 110, and the A / D converter 11 converts the luminance of each pixel into, for example, 8-bit digital data. The shading correction circuit 12 outputs
The image data a00 from the D converter 11 is sent to the line sensor 1
The correction is made based on the sensitivity unevenness of 10 and the illumination unevenness of the lamp 105. The line buffer 13 stores the image data b00 corrected by the shading correction circuit 12 in line units corresponding to the main scanning direction.

As shown in FIG. 3, for example, as shown in FIG.
In the case of reading by i), data for 10 pixels in the sub-scanning direction Y and about 4900 pixels in the main scanning direction X are stored. In an area corresponding to each pixel, 8-bit data b00 representing luminance is stored.

The attribute judging unit 123
Is determined for the image data b00 stored in. The reference selecting unit 1 included in the attribute determining unit 123 selects a reference block 2R including a plurality of pixels constituting an image, and selects an attention area including one or a plurality of pixels in the reference block 2R. The pixels constituting the image here are the pixels whose data is stored in the line buffer 13. The reference block 2R is composed of 4 × 4 16 pixels as shown in FIGS. 3 and 4A, for example. As the region of interest, for example, the pixel of interest b2 in the reference block is selected. The target pixel b2 is a pixel at a predetermined position in the reference block 2R. The reference selecting unit 1 notifies the orthogonal transforming unit 2 of the selected reference block 2R and target pixel b2.

The orthogonal transformation means 2 reads out the data of each pixel in the notified reference block 2R from the line buffer 13 and performs orthogonal transformation on a block basis. Orthogonal data, which is the result of the orthogonal transformation, represents the coefficient of each component in the spatial frequency domain. As shown in FIG. 4A, data a1 to a4 of each pixel in a reference block 2R composed of 4 × 4 16 pixels.
When d4 is given, the same number of orthogonal data K1 to N4 as the number of pixels is obtained. When the image data is represented by 8 bits, for example, truncation or shift of the lower 2 bits does not affect the attribute determination.

As the orthogonal transform, an orthogonal wavelet transform can be used. In this case, for example, the conversion target is 1 ×
4 is given by the data string (A, B, C, D)
First, the result of adding each adjacent element in the first half,
Perform the operation of rearranging the result of the subtraction in the latter half. By this operation, a data string of (A + B, C + D, AB, CD) is obtained. Further, the above operation is performed only for the first half of the data string. by this,
((A + B) + (C + D), (A + B)-(C + D),
AB, CD) are obtained. Orthogonal data (K, L, M,
N) is represented by this data sequence. That is, K
= (A + B) + (C + D), and L = (A + B)-
(C + D), M = AB, and N = CD.

When the data string is represented by an array of two or more rows, the same operation may be performed both in the horizontal direction (row direction) and in the vertical direction (column direction). For example, after an operation is performed in the row direction, an operation is performed in the vertical direction based on the result. It should be noted that, regardless of whether the operation is performed vertically or horizontally,
The end result is the same.

The image data a1 to d4 shown in FIG.
, The orthogonal data K1 to N4 obtained by the wavelet transform are given by the following conversion formulas.

(First column) K1 = (a1 + a2 + a3 + a4 + b1 + b2 + b3 + b4) + (c1 + c2 + c3 + c4 + d1 + d2 + d3
+ d4) K2 = (a1 + a2 + a3 + a4 + b1 + b2 + b3 + b4)-(c1 + c2 + c3 + c4 + d1 + d2 + d3
+ d4) K3 = (a1 + b1 + c1 + d1)-(a2 + b2 + c2 + d2) K4 = (a3 + b3 + c3 + d3)-(a4 + b4 + c4 + d4) (second column) L1 = (((a1 + b1)-(c1 + d1)) + ((a2 + b2)-(c2 + d2))) + (((a3 + b3)
-(c3 + d3)) + ((a4 + b4)-(c4 + d4))) L2 = (((a1 + b1)-(c1 + d1)) + ((a2 + b2)-(c2 + d2 )))-(((a3 + b3)
-(c3 + d3)) + ((a4 + b4)-(c4 + d4))) L3 = ((a1 + b1)-(c1 + d1))-((a2 + b2)-(c2 + d2) ) L4 = ((a3 + b3)-(c3 + d3))-((a4 + b4)-(c4 + d4)) (third column) M1 = ((a1-b1) + (a2-b2)) + ((a3-b3) + (a4-b4)) M2 = ((a1-b1) + (a2-b2))-((a3-b3) + (a4-b4)) M3 = (a1-b1) -(a2-b2) M4 = (a3-b3)-(a4-b4) (4th column) N1 = ((c1-d1) + (c2-d2)) + ((c3-d3) + (c4- d4)) N2 = ((c1-d1) + (c2-d2))-((c3-d3) + (c4-d4)) N3 = (c1-d1)-(c2-d2) N4 = (c3- d3)-(c4-d4) In this case, the orthogonal transform means 2 can be realized by a wavelet transform circuit 14 as shown in FIG.
As shown in FIG. 5, the wavelet transform circuit 14 includes transform operation circuits 1400, 1401-1
414 and 1415. Each conversion operation circuit 14
Since each of 00 to 1415 is composed of only an adder / subtractor, as is clear from each conversion formula, it is possible to realize a high operation speed.

It should be noted that, instead of using such a wavelet transform for the orthogonal transform, other orthogonal transforms such as FFT and DCT may be used.

The orthogonal data e00 as described above is output to the feature calculation means 3.

The characteristic calculating means 3 calculates characteristic values representing the characteristics of each of the low spatial frequency band, the intermediate spatial frequency band, and the high spatial frequency band based on the orthogonal data e00 from the orthogonal transforming means 2. .

The orthogonal data K1 to N4 shown in FIG. 4B corresponding to a reference block consisting of 4 × 4 16 pixels are
For example, as shown in FIG. 6, the data is classified into each spatial frequency band. The orthogonal data K1 is classified into the lowest spatial frequency band. Also, the orthogonal data L1 and K2 are classified into a low spatial frequency band. Also, the orthogonal data K3, L2, M1, N
1, M2, L3, K4, L4, and N2 are classified into intermediate spatial frequency bands. Also, orthogonal data M3, N3, M
4, N4 are classified into high spatial frequency bands. Among them, the orthogonal data K1 classified into the lowest spatial frequency band
Represents a DC component (constant term), and the orthogonal data L1 to N4 classified into other spatial frequency bands represent an AC component. Since the feature of the reference block appears in the AC component, here, the orthogonal data K classified into the lowest spatial frequency band is used.
1 is not used.

In general, a reference block belonging to a character area represented by a line such as a character is characterized in that the absolute value of orthogonal data classified into a low spatial frequency band is large. FIG. 7A shows a specific example of image data when a document is a sheet on which characters are printed. As shown in FIG. 7A, the value of the pixel in the center two columns is considerably larger than the value of the pixel in the other two columns, and this represents a vertical line. FIG. 7B shows a specific example of orthogonal data in this case. As shown in FIG. 7B, except for the orthogonal data classified into the lowest spatial frequency band, the absolute value of the orthogonal data classified into the low spatial frequency band is the absolute value of the orthogonal data classified into another band. Considerably larger than the value.

In a reference block belonging to a halftone dot region having a low resolution, the absolute value of orthogonal data classified into an intermediate spatial frequency band is large. The absolute value of the orthogonal data to be classified increases. FIG. 8A shows a specific example of image data in a case where the original is a dot-printed sheet. FIG. 8B shows a specific example of orthogonal data in this case. As shown in FIG. 8B, in this example, the absolute value of the orthogonal data classified into the intermediate spatial frequency band is considerably large.

Further, in the reference block belonging to the continuous tone area, the tendency that the absolute value of the orthogonal data classified into the specific band becomes large is small. FIG. 9A shows a specific example of image data when the original is photographic paper on which a photographic image is printed. FIG. 9B shows a specific example of orthogonal data in this case. As shown in FIG. 9B, in other cases, orthogonal data having a value exceeding 1000 existed. However, in this case, the orthogonal data did not exist. The value does not change.

Thus, for example, based on the orthogonal data L1 to N4 shown in FIG.
m, the characteristic value Fn of the intermediate spatial frequency band, and the characteristic value Fp of the high spatial frequency band are given by the following equations.

Fm = | L1 | + | K2 | + | L2 | Fn = | M1 | + | N1 | + | M2 | + | N2 | + | K3 | + | L3
| + | K4 | + | L4 | Fp = | M3 | + | N3 | + | M4 | + | N4 | (where | ** | means an absolute value.) In this case, the feature calculating means 3 This can be realized by the calculation circuit 15 as shown in FIG. Calculation circuit 15
Is an absolute value operation circuit 1500a, as shown in FIG.
1500b to 1500n and 1500o. The orthogonal data L1 to N4 are input to the absolute value calculation circuits 1500a to 1500o, respectively. Absolute value calculation circuit 15
00a to 1500o calculate the absolute value of each orthogonal data and adders 1501a, 1501c, 1501d, 15
01e, 1501f, 1501j, and 1501k. For example, each of the absolute value calculation circuits 1500a and 1500b to which the orthogonal data L1 and K2 are input has an absolute value | L1 |
| K2 | is output to the adder 1501a.

The adders 1501a to 1501l are arranged according to the above equation. The adder 1501a is provided with | L1 | from the absolute value arithmetic circuit 1500a and the absolute value arithmetic circuit 150.
| K2 | from 0b is added, and | L1 | + | K2 |
Is output to the adder 1501b. | L1 | + | K2 | from the adder 1501a and | L2 | from the absolute value calculation circuit 1500c are input to the adder 1501b.
The adder 1501b adds these and adds | L1 | + | K
2 | + | L2 |, and the characteristic value Fm of the low spatial frequency band
Is output. Similarly, the adder 1501i outputs the feature value Fn of the intermediate spatial frequency band, and the adder 1501l outputs
The feature value Fp of the low spatial frequency band is output.

FIGS. 7 (b), 8 (b), 9
When each feature value is calculated based on each orthogonal data shown in (b), Fm = 1564 for the example of FIG.
The result is Fn = 1356 and Fp = 334.
Similarly, for the example of FIG. 8B, Fm = 1163,
The result is that Fn = 808 and Fp = 15. Also, in the example of FIG. 9B, Fm = 148, Fn =
The result is 100, Fp = 10.

These characteristic values Fm, Fn, Fp are output to the attribute determining means 4.

The attribute determining means 4 determines the attribute of the target pixel b2 based on each characteristic value. The attribute determining means 4 can be realized by, for example, a determining circuit 16 as shown in FIG. As shown in FIG. 11, the determination circuit 16 includes comparators 1600a, 1600b, and 1601.
The comparator 1600a compares the magnitude of the feature value Fm with the magnitude of the feature value Fn, and outputs the larger value. Comparator 1600b
Compares the magnitude of the output of the comparator 1600a with the feature value Fp and outputs the larger value. Comparator 1601 compares the magnitude of output Fmax of comparator 1600b with threshold value T1.

The threshold value T1 is set in advance according to the size of the number of bits representing the image data of one pixel and the size of the number of pixels of the reference block. If one pixel of image data is represented by 8 bits and the number of pixels of the reference block is 4 × 4, that is, 16 pixels, it is set to about “500”.

If Fmax <T1, predetermined data indicating the continuous tone area is output as the determination result Q _ans for the target pixel b2. When a 2-bit logical signal is used to represent the determination result Q _ans , for example, the determination result Q _ans indicating a continuous tone area is represented by “00”. If Fmax <T1 is not satisfied, Fmax is determined according to which of Fm, Fn, and Fp the feature value is.
Is a feature value Fm, predetermined data indicating a character area, for example, “11” is output as the determination result Q _ans for the target pixel b2. Also, Fmax is the feature value Fn
Alternatively, in the case of Fp, predetermined data indicating a halftone dot area, for example, “10” is output as the determination result Q _ans for the target pixel b2.

When the attribute of the pixel of interest b2 in the reference block 2R is determined in this way, the attribute determination unit 4 notifies the reference selection unit 1 of the determination. Upon receiving this notification, the reference selecting means 1 selects the next reference block 2R and target pixel b2. The position of the next reference block 2R is obtained by shifting the position of the preceding reference block 2R by at least a predetermined number of pixels in at least one of the main scanning direction X and the sub-scanning direction Y.

For example, the position of the reference block 2R is moved one pixel at a time in the main scanning direction X to sequentially select the reference block 2R. When the selection in the main scanning direction X is completed, the reference block 2R is moved by one pixel in the sub-scanning direction Y next. Then, the main scanning direction X is selected again. The arrangement of the pixel of interest b2 with respect to the reference block 2R does not change. If the above-described attribute determination is repeated each time selection is performed, attribute determination can be performed for all data b00 stored in the line buffer 13.

Although an example has been described in which the reference block 2R is 4 × 4 16 pixels, the amount of information obtained from the image data corresponding to the reference block 2R is small in this case. The greater the number of pixels of the reference block 2R, the higher the accuracy of attribute determination. When the wavelet transform is used, the number of pixels of the reference block 2R in the main scanning direction X and the sub-scanning direction Y needs to take a value of a power of two.
Therefore, practically, 8 × 8 is used as the reference block 2R.
It is preferable to use a pixel including 64 or more pixels.

However, in order to increase the number of pixels in the sub-scanning direction Y of the reference block 2R, the number of lines in the line buffer 13 must be ensured accordingly. From the viewpoint of cost, the number of pixels in the sub-scanning direction Y is preferably set to about four pixels. On the other hand, even if the number of pixels in the main scanning direction X is set large, there is no need to increase the number of lines in the line buffer 13. Therefore, instead of using a reference block including 8 × 8 64 pixels, a reference block including 4 × 16 64 pixels may be used.

Further, depending on the position of the reference block 2R selected by the reference selecting means 1, there may be a case where pixels constituting an image are included only in a partial area within the reference block 2R.

For example, as shown in FIG. 12, when the pixel at the position corresponding to the corner of the image data b00 stored in the line buffer 13 is selected as the pixel of interest b2, the pixels a1, b1, c1, c1,. d1, a2, a3,
The data of a4 is not stored in the line buffer 13.

In such a case, the pixels a1, b1, c1, d1, and a1 other than the pixels whose data b00 is stored in the line buffer 13 among the pixels in the reference block 2R.
The values of 2, a3 and a4 are filled with values representing a white image. For example, when the image data is represented by 8 bits, “0”, “1”,..., “254”,
The symbol “255” can be assigned. In this case, data “0” is a value representing a white image. Note that, contrary to this example, for each gradation from black to white,
When the symbols “0”, “1”,..., “254”, and “255” are assigned, the value representing the white image is “255”.

Filling with a value representing a white image as described above generally occurs in most cases where white, which is the background color of the document, appears at the edge of the document image, and the white image influences the attribute determination. Because it has no effect.

As a result, even when the pixel data in the reference block is partially missing, the attribute can be determined for each pixel.

The result of the determination by the attribute determining means 4 is output from the attribute determining section 123 to the image processing section 124.

The image processing section 124 performs the image processing corresponding to each of the halftone area, the continuous tone area, and the character area.
6. A character image processing unit 127 is provided.

Under the control of the control unit 128, the dot image processing unit 125, the photograph image processing unit 126, and the character image processing unit 127 read the image data b00 stored in the line buffer 13. For the read image data b00, for example, the character image processing unit 127
A simple binarization process is performed. Further, the image processing unit for photograph 12
6. The dot image processing unit 125 performs an error diffusion process and a dither process. Thereby, a gray scale is expressed in a pseudo manner.
In order to improve the quality of the reproduced image, the halftone dot image processing unit 125 may perform moiré removal processing and the character image processing unit 127 may perform sharpening processing as necessary.

The outputs of the dot image processing unit 125, the photograph image processing unit 126, and the character image processing unit 127 are as follows:
It is input to the selector section 129. The selector unit 129 is
5 is a specific example of the image processing selection unit 7. The control unit 128
According to the determination result from the attribute determination unit 123, the selector unit 1
29 causes any input to be selected. For example, if the determination result is “00”, the selector unit 129 outputs an input from the photographic image processing unit 126 to the printer unit 102. If the determination result is “11”, the selector unit 1
Reference numeral 29 outputs the input from the character image processing unit 127 to the printer unit 102. If the determination result is “10”, the selector unit 129 outputs an input from the dot image processing unit 125 to the printer unit 102.

As a result, the image processing is performed according to the determination result for each pixel, so that an image can be recorded on a sheet with high image quality. (Second Embodiment) FIG. 13 shows a schematic configuration of a signal processing unit 111 according to a second embodiment.

As shown in FIG. 13, the signal processing unit 11
Reference numeral 1 denotes an arrangement in which an integrating means 5 is arranged between the feature calculating means 3 and the attribute determining means 4 of the signal processing unit 111 in the first embodiment. Since the operation is the same as that already described in the first embodiment, a detailed description of the operation from the line sensor 110 to the feature calculation unit 3 is omitted, but in the image processing method, as shown in FIG. First, the reference selecting means 1 selects the reference block 2R and the target pixel b2 (S1). Next, the orthogonal transformation means 2 sets the reference block 2
The value of each pixel in R is orthogonally transformed to obtain orthogonal data (S
2). Next, the feature calculation means 3 calculates feature values representing the features of each of the low spatial frequency band, the intermediate spatial frequency band, and the high spatial frequency band (S3).

Here, the reference block 2 including the target pixel g
For R, the characteristic of the low spatial frequency band is Fm (g),
It is assumed that the characteristic value of the intermediate spatial frequency band is given by Fn (g) and the characteristic value of the high spatial frequency band is given by Fp (g).

Next, the integrating means 5 integrates the characteristic value of the reference block 2R including the target pixel g and the characteristic value of another block for each spatial frequency band (S4). For this purpose, the integrating means 5 also needs, for example, a feature value of a block in which pixels (g−1) and (g + 1) adjacent to the target pixel g on both sides in the main scanning direction X are the target pixels. Reference block 2R including target pixel (g-1), reference block 2 including target pixel g
When the characteristic value is obtained in the order of R and the reference block 2R including the target pixel (g + 1), the target pixel g, (g−1) is obtained until the characteristic value of the reference block 2R including the target pixel (g + 1) is obtained. Is stored in a register or the like.

For the block including the target pixels (g-1) and (g + 1), the characteristics of the low spatial frequency band are Fm (g-1) and Fm (g + 1), respectively, and the characteristic value of the intermediate spatial frequency band Are Fn (g-1) and Fn (g + 1), and the feature values in the high spatial frequency band are Fp (g-1) and Fp (g +
Suppose that it was given in 1).

The integrating means 5 integrates these characteristic values for each spatial frequency band. That is, the integrated value Fms of the characteristic value
(G), Fns (g) and Fps (g) are given by the following equations.

Fms (g) = Fm (g−1) + Fm (g) + Fm (g
+1) Fns (g) = Fn (g-1) + Fn (g) + Fn (g
+1) Fps (g) = Fp (g-1) + Fp (g) + Fp (g
+1) In this example, the reference block 2R including the target pixel g
The feature values of the block having pixels on both sides in the main scanning direction X as the pixel of interest are integrated with respect to the feature value of, but the present invention is not limited to this. For example, pixels around the target pixel g,..., (G−2), (g−1), and a pixel (g +
As in 1), (g + 2),..., The feature values of blocks including a predetermined number of consecutive pixels arranged one by one in the main scanning direction X may be integrated. The pixel (g-
4), (g−2), and pixel (g + 2) and (g + 4), respectively, are integrated with the feature values of a block including a predetermined number of non-consecutive pixels arranged one by one in the main scanning direction X. You may. Of course, the arrangement direction of each target pixel is not limited to the main scanning direction X. Although a relatively large number of data are stored in the line buffer 13 in the main scanning direction X, it is preferable that the arrangement direction is the sub-scanning direction Y or an oblique direction between the main scanning direction X and the sub-scanning direction Y. .

The integrating means 5 can be realized by an integrating circuit 17 as shown in FIG. 15, for example. The integrating circuit 17 includes registers 1701a to 1701f and adders 1700a to 1700f. In the integration circuit 17, the configuration for integrating the characteristic values for each spatial frequency band is the same. Therefore, here, the configuration for the characteristic value Fm of the low spatial frequency band will be described. The registers 1701a and 1701d of the integrating circuit 17 are shift registers, and the register 1701d always stores the feature value Fm (g−g−g) of the reference block 2R including the target pixel (g−1) immediately before the target pixel g. 1) is in a latched state. Further, the register 1701a is in a state where the characteristic value Fm (g) of the reference block 2R including the target pixel g is always latched. In this state, when the feature value Fm (g + 1) of the reference block 2R including the target pixel (g + 1) following the target pixel g is input, the adder 1700 is added.
a is the input feature value Fm (g + 1) and the register 17
01a is added to the characteristic value Fm (g) held in 01a. The adder 1700d calculates the addition result of the adder 1700a and the characteristic value Fm (g) held in the register 1701d.
-1) is added. Thereby, the adder 1700a
Fm (g-1) + Fm (g) + Fm (g + 1)
= Fms (g). Similarly, an adder 1700e
Output of Fn (g-1) + Fn (g) + Fn (g +
1) = Fns (g) is obtained. The adder 1700
The output of f includes Fp (g-1) + Fp (g) + Fp (g
+1) = Fps (g) is obtained.

When the integration is performed by the integration means 5, the characteristic value of the reference block may not exist depending on the position of the reference block in which the characteristic value is integrated. For example, there is a case where the target pixel is located outside the image, such as when the data of the target pixel is not stored in the line buffer 13. No feature value has been calculated for the reference block. In this case, the integrating means 5 only needs to set each characteristic value of the reference block to “0” and integrate only the characteristic values of the reference block having the target pixel in the image.

Next, the attribute determining means 4 determines the attribute of the target pixel g based on the integration result of the integrating means 5 (S
5). The attribute determining means 4 in this case can be realized by, for example, a determining circuit 18 as shown in FIG.

The determination circuit 18 has the same configuration as the determination circuit 16 and includes comparators 1800a, 1800b, and 1801. The comparator 1800a calculates the integration result Fms
(G) is compared with the integration result Fns (g), and the larger value is output. The comparator 1800b includes a comparator 18
The output 00a is compared with the integration result Fps (g), and the larger value is output. Comparator 1801 compares the magnitude of output Fsmax of comparator 1800b with threshold value T2.

The threshold value T2 is determined by the size of the number of bits representing the image data of one pixel, the size of the number of pixels of the reference block,
It is set in advance corresponding to the number of feature values to be integrated. If the image data of one pixel is represented by 8 bits and the number of pixels of the reference block is 4 × 4 = 16 pixels, “5
00 ”× 3 = about“ 1500 ”.

If Fsmax <T2, predetermined data indicating the continuous tone area is output as the determination result Q _ans for the target pixel g. When a 2-bit logical signal is used to represent the determination result Q _ans , for example, the determination result Q _ans indicating a continuous tone area is represented by “00”. If Fsmax <T2, Fsmax is a characteristic value Fms (g) according to which of Fs (g), Fns (g), and Fps (g) Fsmax is.
In this case, predetermined data indicating a character area, for example, “11” is output as the determination result Q _ans for the target pixel g. When Fsmax is the feature value Fns (g) or Fps (g), predetermined data indicating a halftone area, for example, “10” is output as the determination result Q _ans for the target pixel g.

As described above, by integrating the characteristic values of the other blocks having pixels around the target pixel with the target pixel as the target pixel with the characteristic values of the reference block and performing attribute determination based on the result of the integration, a more detailed attribute can be obtained. Judgment becomes possible. (Third Embodiment) Although the signal processing unit 111 in the first embodiment can perform attribute determination in more detail than in the past, halftone dots with low resolution and characters or lines with high resolution can be used. May have similar features. In this case, an erroneous determination may be made.

FIG. 20A shows a specific example of image data when the original is a sheet on which halftone printing with a low resolution has been performed. FIG. 20B shows a specific example of orthogonal data in this case. As shown in FIG. 20 (b), the absolute value of orthogonal data classified into the intermediate spatial frequency band, for example, the value in the second row and third column and the value in the second row and fourth column is slightly compared with the absolute values of other orthogonal data. Although it is large, it does not show a particularly large value.

On the other hand, FIG. 21A shows a specific example of image data when the original is a sheet on which high-resolution characters and lines are printed. FIG. 21B shows a specific example of orthogonal data in this case. As shown in FIG. 21B, the absolute value of the orthogonal data classified into the low spatial frequency band, for example, the value in the first row and the second column is slightly larger,
The absolute value of the orthogonal data classified into the intermediate spatial frequency band,
For example, the value in the first row and the fourth column is slightly larger.

That is, in the case of a character or a line having a high resolution,
Both the characteristics of characters and lines and the characteristics of halftone dots appear, which causes erroneous determination.

For example, when the characteristic value is calculated in the same manner as in the first embodiment, in the example of FIG.
134, Fn = 1358, and Fp = 130. In the example of FIG. 21B, Fm = 9
14, Fn = 954 and Fp = 10 are obtained.

In any case, the feature value Fn of the intermediate spatial frequency band is maximum, and the result of attribute determination in both cases indicates a halftone dot area.

By the way, generally, in the case of a vertical straight line or a horizontal straight line, there is a feature that the absolute value of specific orthogonal data among the orthogonal data becomes large.

For example, the orthogonal data K1 shown in FIG.
In the case of ~ N4, when there is a vertical straight line, the orthogonal data L1,
There is a feature that the absolute values of M1 and N1 increase. Also, when there is a horizontal straight line, the absolute value of the orthogonal data K2, K3, and K4 is large.

Therefore, as shown in FIG. 17, the signal processing unit 111 according to the third embodiment includes a correction calculation unit 6 in addition to the configuration of the signal processing unit 111 according to the first embodiment. Note that the description of the operation of the signal processing unit 111 from the A / D converter 11 to the feature calculation unit 3 is the same as the operation described in the first embodiment, and a description thereof will be omitted.

The correction calculating means 6 calculates a correction value based on orthogonal data of a predetermined spatial frequency band among the orthogonal data. For example, orthogonal data of a predetermined spatial frequency band is
These are orthogonal data L1, M1, and N1 related to a vertical straight line, and orthogonal data K2, K3, and K4 related to a horizontal straight line.
In this case, the correction calculation unit 6 can be realized by, for example, a correction circuit 19 as shown in FIG.

As shown in FIG. 18, the correction circuit 19 includes absolute value calculation circuits 1900a to 1900f. The orthogonal data L1, M1, N1, K2, K3, and K4 are input to the absolute value calculation circuits 1900a to 1900f, respectively. Absolute value calculation circuits 1900a to 1900f calculate the absolute value of each orthogonal data and adders 1901a and 1901a
1c. For example, the absolute value calculation circuits 1900a and 1900b to which the orthogonal data L1 and M1 are input output the absolute values | L1 | and | M1 | to the adder 1901a.

The adder 1901a is connected to the absolute value arithmetic circuit 19
| L1 | from 00a and | M1 | from the absolute value calculation circuit 1900b are added, and | L1 | + | M1 | is output to the adder 1901b. | L1 | + | M1 | from the adder 1901a and | N1 | from the absolute value calculation circuit 1900c are input to the adder 1901b. Adder 15
01b is the sum of these | L1 | + | M1 | + | N
1 | is obtained and output to the adder 1901e.

The adder 1901c is provided with | K2 | from the absolute value operation circuit 1900d and the absolute value operation circuit 1900d.
| K3 | from E, and outputs | K2 | + | K3 | to the adder 1901d. In the adder 1901d,
| K2 | + | K3 | from the adder 1901c and | K4 | from the absolute value calculation circuit 1900f are input. The adder 1901d adds these and adds | K2 | + | K3 |
+ | K4 | is obtained and output to the adder 1901e.

The adder 1901e is provided with | L1 | + | M1 | + | N1 | from the adder 1901b and the adder 1901d.
| K2 | + | K3 | + | K4 | from the data, and | L1 | + | M1 | + | N
1 | + | K2 | + | K3 | + | K4 |
Output to

The attribute determining means 4 determines the attribute of the target pixel b2 based on the characteristic value and the correction value. In this case, the attribute determining means 4 is, for example, a determination circuit 20 as shown in FIG.
It can be realized by.

As shown in FIG. 19, the judgment circuit 20 includes comparators 2000a, 2000b,
001, a comparator 2002, a selector circuit 20
03. The comparator 2000a compares the magnitude of the feature value Fm with the magnitude of the feature value Fn, and outputs the larger value. The comparator 2000b compares the magnitude of the output of the comparator 2000a with the feature value Fp, and outputs the larger value.
Comparator 2001 compares the magnitude of output Fmax of comparator 2000b with threshold value T1.

If Fmax <T1, predetermined data indicating the continuous tone area is output to the selector circuit 2003 as the determination result Q _ans for the target pixel b2.
When a 2-bit logical signal is used to represent the determination result Q _ans , for example, the determination result Q _ans indicating a continuous tone area is represented by “00”. If Fmax <T1 is not satisfied, predetermined data indicating a character area when Fmax is the feature value Fm, for example, “11”, is determined according to which of Fmax is the feature value Fm, Fn, or Fp. To the selector circuit 2003 as the determination result Q _ans for the pixel of interest b2. Fmax is the characteristic value Fn or F
In the case of p, predetermined data indicating a halftone area,
For example, the determination result Q _ans for the pixel of interest b2 is “10”.
Is output to the selector circuit 2003.

Further, in the determination circuit 20, the comparator 2
002 compares the magnitude of the correction value Rq with the threshold T3.
The threshold value T3 is set corresponding to the threshold value T1, and for example, when the threshold value T1 is about “500”, “500” × 2 =
It is set to about “1000”. And the comparator 200
2, when Rq <T3, the selector circuit 20 uses predetermined data indicating that the area is not a character area as the correction result R _ans.
Output to 03. If the correction result using the 2-bit logic signal to represent R _ans, the correction result R _ans indicating that it is not, for example, the character area represented by "10". Also, Rq <
If it is not T3, predetermined data indicating a character area, for example, “11” is output to the selector circuit 2003 as a correction result R _ans .

The selector circuit 2003 determines the determination result Q _ans
When it is "10" indicating the dot region, the correction according to the result R _ans, and outputs the judgment result Q _ans' corrected by the decision result Q _ans or correction result R _ans. When the judgment result Q _ans matches the correction result R _ans , the judgment result Q _ans is output. On the other hand, when the determination result Q _ans is different from the correction result R _ans , that is, when the correction result R _ans is “11” indicating a character area, the correction result R _ans is output as the determination result Q _ans ′.
If the determination result Q _ans is not “10” indicating a halftone dot area, the determination result Q _ans is output regardless of the correction result R _ans .

As described above, the provision of the correction calculating means 6 reduces the possibility that a part of a character represented by a vertical or horizontal straight line will become a halftone dot, and makes it possible to obtain favorable attributes for the halftone dot region and the character region. The determination can be realized. (Fourth Embodiment) As shown in FIG. 22, a signal processing unit 111 according to a fourth embodiment includes a correction calculation unit 6 in addition to the configuration of the signal processing unit 111 according to the second embodiment. Prepare. The signal processing unit 1 from the A / D converter 11 to the feature calculation unit 3 and the correction calculation unit 6
The description of the operation No. 11 is the same as the operation described in the first and third embodiments, and will not be repeated.

Here, the reference block 2 including the target pixel h
For R, the characteristic value of the low spatial frequency band is Fm (h)
Suppose that the characteristic value of the intermediate spatial frequency band is given by Fn (h) and the characteristic value of the high spatial frequency band is given by Fp (h). Further, for the reference block 2R having the pixels (h-1) and (h + 1) adjacent to both sides of the pixel of interest h in the main scanning direction X, the feature of the low spatial frequency band is Fm (h- 1), Fm (h + 1), and the feature values of the intermediate spatial frequency band are Fn (h-1), Fn (h + 1)
And the feature values of the high spatial frequency band are Fp (h-1), Fp
Suppose that it was given by (h + 1).

Further, the target pixel h and the target pixel (h-
1) The correction value is Rq for the pixel of interest (h + 1).
(H), Rq (h-1), and Rq (h + 1).

The integrating means 5 integrates these characteristic values for each spatial frequency band. That is, the integrated value Fms of the characteristic value
(H), Fns (h) and Fps (h) are given by the following equations.

Fms (h) = Fm (h−1) + Fm (h) + Fm (h
+1) Fns (h) = Fn (h-1) + Fn (h) + Fn (h
+1) Fps (h) = Fp (h-1) + Fp (h) + Fp (h
+1) The integrating means 5 also integrates each correction value. That is,
The integration result Rqs (h) of the correction values is given by the following equation.

Rqs (h) = Rq (h-1) + Rq
(H) + Rq (h + 1) In this example, the reference block 2R including the target pixel h
Pixels on both sides in the main scanning direction X (h
Although the feature value and the correction value are integrated for the block in which -1) and (h + 1) are the target pixels, the present invention is not limited to this. For example, pixels around the pixel of interest h ..., (h-
2), (h-1) and pixels (h + 1), (h + 2), ...
As described above, the feature value and the correction value of a block including a predetermined number of consecutive pixels arranged one by one in the main scanning direction X may be integrated. Also, pixels ..., (h-4), (h
-2) and the feature value and the correction value of a block including a predetermined number of non-consecutive pixels arranged one by one in the main scanning direction X, such as pixels (h + 2), (h + 4),. You may. Of course, the arrangement direction of each target pixel is not limited to the main scanning direction X. Although a relatively large amount of data is stored in the line buffer 13 in the main scanning direction X, it is preferable.
It may be an oblique direction between the main scanning direction X and the sub-scanning direction Y.

The integrating means 5 in this case can be realized by an integrating circuit 21 as shown in FIG. 23, for example. The integrating circuit 21 includes registers 2101a to 2101
h, adders 2100a to 2100h. In the integrating circuit 21, the configuration for integrating the characteristic values for each spatial frequency band and the configuration for integrating the correction values are the same. Therefore, here, the characteristic value F of the low spatial frequency band is used.
The configuration for m will be described. The registers 2101a and 2101e of the integrating circuit 21 are shift registers. The register 2101e always includes the reference block 2 including the target pixel (h-1) immediately before the target pixel h.
The characteristic value Fm (h-1) of R is in a latched state. The register 2101a always stores the target pixel h
The characteristic value Fm (h) of the reference block 2 </ b> R including the above is latched. In this state, the feature value F of the reference block 2R including the target pixel (h + 1) next to the target pixel h
When m (h + 1) is input, the adder 2100a adds the input characteristic value Fm (h + 1) and the characteristic value Fm (h) held in the register 2101a.
The adder 2100e adds the addition result of the adder 2100a and the feature value Fm (h-1) held in the register 2101e. As a result, the output of the adder 2100e becomes Fm (h-1) + Fm (h) + Fm (h + 1) = Fms
(H) is obtained. Similarly, the output of the adder 2100f includes Fn (h-1) + Fn (h) + Fn (h + 1) = F
ns (h) is obtained. Further, the output of the adder 2100g includes Fp (h-1) + Fp (h) + Fp (h + 1) =
Fps (h) is obtained. Further, the output of the adder 2100h includes Rq (h-1) + Rq (h) + Rq (h +
1) = Rqs (h) is obtained.

When the integration is performed by the integration means 5, the characteristic value of the reference block may not exist depending on the position of the reference block in which the characteristic value and the correction value are integrated. For example, there is a case where the target pixel is located outside the image, such as when the data of the target pixel is not stored in the line buffer 13. No feature value has been calculated for the reference block. In this case, the integrating means 5 only needs to set each characteristic value and correction value of the reference block to “0” and integrate only the characteristic value and the correction value of the reference block having the target pixel in the image.

The attribute determining means 4 determines the attribute of the pixel of interest h based on the result of integration by the integrating means 5. In this case, the attribute determination means 4 is, for example, a determination circuit 2 as shown in FIG.
2 can be realized.

The determination circuit 22 has the same configuration as the determination circuit 18 and includes comparators 2200a, 2200b, 2201,
2202 and a selector circuit 2203. The comparator 2200a calculates the characteristic value Fms (h) and the characteristic value Fns
(H) is compared, and the larger value is output. The comparator 2200b outputs the output of the comparator 2200a and the characteristic value F
Compare the magnitude with ps (h) and output the larger value. The comparator 2201 outputs the output Fsma of the comparator 2200b.
The magnitude of x and the threshold value T2 are compared.

If Fsmax <T2, the continuous tone region
Is determined for the pixel of interest h.
Fixed result Q_ansIs output to the selector circuit 2203.
Judgment result Q_ansUse a 2-bit logic signal to represent
If there is, for example, a determination
Fruit Q_ansIs represented by “00”. Also, Fsmax <T2.
Fsmax is the characteristic value Fms (h), Fns
(H) or Fps (h) according to Fs (h).
When max is the feature value Fms (h), the character area
, For example, “11” is assigned to the pixel of interest h.
Judgment result Q _ansOut to the selector circuit 2203 as
Power. Also, Fsmax is the feature value Fns (h) or Fp
In the case of s (h), a predetermined data indicating a halftone dot area is used.
Data, for example, “10” is determined as the determination result Q for the pixel of interest h.
_ansIs output to the selector circuit 2203.

Further, the comparator 2202 calculates the correction value Rqs
(H) is compared with the threshold value T4. The threshold value T4 is set corresponding to the threshold value T2.
If “0” × 3 = about “1500”, “1500”
× 2 = set to about “3000”. Then, when Rqs (h) <T4, the comparator 2202 outputs, to the selector circuit 2203, predetermined data indicating a non-character area as a correction result R _ans . If the correction result using the 2-bit logic signal to represent R _ans, the correction result R _ans indicating that it is not, for example, the character area represented by "10". If Rqs (h) <T3, predetermined data indicating a character area, for example, “11” is output to the selector circuit 2203 as a correction result R _ans .

The selector circuit 2203 determines the determination result Q _ans
When it is "10" indicating the dot region, the correction according to the result R _ans, and outputs the judgment result Q _ans' corrected by the decision result Q _ans or correction result R _ans. When the judgment result Q _ans matches the correction result R _ans , the judgment result Q _ans is output. On the other hand, when the determination result Q _ans is different from the correction result R _ans , that is, when the correction result R _ans is “11” indicating a character area, the correction result R _ans is output as the determination result Q _ans ′.
If the determination result Q _ans is not “10” indicating a halftone dot area, the determination result Q _ans is output regardless of the correction result R _ans .

As a result, more detailed attribute determination can be performed on the target pixel h of the reference block 2R than in the first, second, and third embodiments. (Others) In the first to fourth embodiments, the present invention is applied to a digital copying machine. However, the present invention is not limited to this. For example, the invention can be applied to other image processing apparatuses such as a scanner and a facsimile.

FIG. 25 shows a schematic configuration of the signal processing section 111 when the present invention is applied to a scanner.

As shown in FIG. 25, the configuration of attribute determining section 123 is the same as the configuration of attribute determining section 123 in the second embodiment. Although not shown, the configuration upstream of the line sensor 110 can be configured similarly to the image scanner unit 101 of the digital copying machine. However,
When applied to a digital copying machine, the attribute determining unit 123
Is used to improve image quality, but when applied to a scanner, it may be used to select a compression format.

The compression processing section 130 selects a compression mode according to the determination result from the attribute determination section 123, compresses the image data read by the image scanner section, and stores it in the data storage section 136.

The compression processing section 130 is a high compression processing section 13
1, a middle compression processing section 132 and a low compression processing section 133.

High compression processing section 131, medium compression processing section 13
2. The low-compression processing unit 133 reads out the image data stored in the line buffer 13 under the control of the control unit 134. The high compression processing unit 131 performs a compression process on the read image data b00 in a compression mode having the highest compression ratio. Further, the middle compression processing section 132 performs the compression processing in a compression mode having the next highest compression ratio. Further, the low-compression processing unit 133 performs the compression processing in a compression mode having the lowest compression ratio.

Here, the compression mode includes at least one of a compression method and a compression ratio. For example, the high compression processing unit 13
1. When different compression methods are used by the middle compression processing unit 132 and the low compression processing unit 133, the high compression processing unit 131
Uses the compression scheme with the highest compression rate among those compression schemes. When the high compression processing unit 131, the medium compression processing unit 132, and the low compression processing unit 133 use the same compression method with a variable compression ratio, the high compression processing unit 131 uses the highest compression method of the compression method. Perform compression at a rate.

High compression processing section 131, medium compression processing section 13
2. Each output of the low-compression processing unit 133 is input to the selector unit 135. The selector unit 135 is also a specific example of the image processing selection unit 7. The control unit 134 controls the attribute determination unit 1
In accordance with the determination result from 23, the selector unit 135 is made to select any one of the inputs.

For example, if the determination result is “00”, the selector unit 135 stores the input from the low compression processing unit 133 in the data storage unit 136. In this case, it is important to suppress image quality degradation during compression. If the determination result is “11”, the selector unit 135 stores the input from the high compression processing unit 131 in the data storage unit 136.
If the object is a character, it is possible to read even if the image quality is slightly degraded, and in this case, reduction of the consumed capacity is regarded as important. If the determination result is “10”, the selector unit 135
Transmits the input from the medium compression processing unit 132 to the data storage unit 13
6 is stored. In this case, reduction of image quality degradation and reduction of consumption capacity are regarded as equally important.

As described above, the image quality and the compression ratio according to the attribute can be selected with high accuracy.

The present invention can be applied not only to an image processing apparatus for reading an image of a document, such as a digital copying machine or a scanner, but also to an image processing apparatus such as a digital still camera.

Furthermore, in each of the first to fourth embodiments,
The attribute is determined for the target pixel, but the present invention is not limited to this. Instead of the target pixel, the attribute determination may be performed on a target region including a plurality of pixels in the reference block. In this case, although the accuracy is slightly deteriorated as compared with the case where the attribute determination is performed on the target pixel, the processing amount of the attribute determination can be reduced.

Further, in each of the first to fourth embodiments, the signal processing section 111 is constituted by the dedicated circuit, but the present invention is not limited to this. For example, it is also possible to cause the microprocessor to execute the procedure shown in FIG. 14 according to control of a program. In this case, the program may be stored in a non-volatile memory or the like which can be read by the microprocessor, and the microprocessor may read the program when an image is input. The program realizes the attribute determination function and the image processing function as described above in cooperation with the microprocessor and the line buffer.

Such an image processing program is distributed in a state of being stored in a non-volatile memory or the like incorporated in a device, distributed in a state of being recorded on a computer-readable recording medium such as a CD-ROM, or distributed on the Internet. It may be distributed through telecommunication lines such as

The present invention can be applied to an image processing apparatus that handles color images and an image processing apparatus that handles only monochrome images.

[0120]

As described above, according to the present invention, attributes such as a halftone dot region, a character region, and a continuous tone region are determined for each region of interest based on the characteristic value of the reference block. Can be improved. As a result, an appropriate image processing can be selected when an image is recorded on a sheet, and an appropriate compression mode can be selected when an image is compressed. Of course, processing such as the conventional inverse conversion is unnecessary, and the processing time is short.

If the characteristic value of the reference block and the characteristic value of the other blocks are integrated for each spatial frequency, the attribute determination accuracy can be improved by that amount. Since the characteristic values of the other blocks are integrated and the attribute determination is performed only for the attention area, the adverse effect of the erroneous determination can be suppressed.

Further, by using a correction value based on orthogonal data of a predetermined spatial frequency band among the orthogonal data for attribute determination, for example, a characteristic value may be similar between a halftone dot region and a character region. However, erroneous determination in that case can be suppressed.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of a signal processing unit provided in a digital copying machine according to a first embodiment;

FIG. 2 is a diagram for explaining the overall configuration of a digital copying machine to which the present invention has been applied;

FIG. 3 is a diagram for explaining a configuration of a reference block and a state of selection of a reference block;

FIG. 4 is a diagram for explaining a relationship between image data and orthogonal data.

FIG. 5 is a diagram showing a configuration of a wavelet transform circuit.

FIG. 6 is a diagram showing classification based on a spatial frequency band in a reference block.

FIG. 7 is a diagram illustrating a specific example of image data and orthogonal data when a document is a sheet on which characters are printed;

FIG. 8 is a diagram showing a specific example of image data and orthogonal data when a document is halftone printed paper;

FIG. 9 is a diagram showing a specific example of image data and orthogonal data when a document is photographic paper on which a photographic image is printed.

FIG. 10 illustrates a configuration of a calculation circuit.

FIG. 11 is a diagram illustrating a configuration of a determination circuit according to the first embodiment;

FIG. 12 is a diagram showing a state where a reference block including a pixel in which no data is stored is selected;

FIG. 13 is a block diagram illustrating a schematic configuration of a signal processing unit provided in a digital copying machine according to a second embodiment;

FIG. 14 is a flowchart illustrating an image processing method according to a second embodiment.

FIG. 15 is a diagram illustrating a configuration of an integrating circuit according to a second embodiment;

FIG. 16 is a diagram illustrating a configuration of a determination circuit according to the second embodiment.

FIG. 17 is a block diagram illustrating a schematic configuration of a signal processing unit provided in a digital copying machine according to a third embodiment;

FIG. 18 is a diagram illustrating a configuration of a correction circuit according to a third embodiment.

FIG. 19 is a diagram illustrating a configuration of a determination circuit according to a third embodiment;

FIG. 20 is a diagram showing a specific example of image data and orthogonal data for a halftone dot region having a low resolution.

FIG. 21 is a diagram showing a specific example of image data and orthogonal data for a high-resolution character (straight line) area;

FIG. 22 is a block diagram illustrating a schematic configuration of a signal processing unit provided in a digital copying machine according to a fourth embodiment;

FIG. 23 is a diagram illustrating a configuration of an integrating circuit according to a fourth embodiment.

FIG. 24 is a diagram illustrating a configuration of a determination circuit according to a fourth embodiment.

FIG. 25 is a block diagram illustrating a schematic configuration of a signal processing unit when the present invention is applied to a scanner.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Reference selection means 2 Orthogonal transformation means 2R Reference block 3 Feature calculation means 4 Attribute determination means 5 Accumulation means 6 Correction calculation means 7 Image processing selection means 11 A / D converter 12 Shading correction circuit 13 Line buffer 101 Image scanner unit 102 Printer Unit 104 document 110 line sensor s00 image signal a00 image data b00 image data b2 pixel of interest e00 orthogonal data a1 to d4 image data Fm feature value Fn feature value Fp feature value

Continued on the front page (72) Inventor Hideyuki Kuwano 1006 Kazuma Kadoma, Osaka Pref. Matsushita Electric Industrial Co., Ltd. 5C077 LL19 MP01 PP27 PP49 PQ24 5L096 FA21 GA19 JA11

Claims (15)

[Claims] 1. A reference selecting means for selecting a reference block including a plurality of pixels constituting an image, and selecting a region of interest comprising one or more pixels in the reference block; Orthogonal transform means for obtaining orthogonal data which is a result of orthogonally transforming a value; feature calculating means for calculating a feature value representing a feature for each spatial frequency band of the reference block based on the orthogonal data; And an attribute determining means for determining an attribute of the attention area. 2. A reference selecting unit for selecting a reference block including a plurality of pixels constituting an image and selecting an attention area including one or more pixels in the reference block, and Orthogonal transform means for obtaining orthogonal data as a result of orthogonally transforming a value; feature calculating means for calculating a feature value representing a feature for each spatial frequency band of the reference block based on the orthogonal data; and features of the reference block. An image processing apparatus comprising: integrating means for integrating a value and a characteristic value of another block for each spatial frequency band; and attribute determining means for determining an attribute of the attention area based on a result of integration by the integrating means. 3. A reference selecting means for selecting a reference block including a plurality of pixels constituting an image, and selecting an attention area consisting of one or more pixels in the reference block; Orthogonal transformation means for obtaining orthogonal data that is a result of orthogonal transformation of a value; feature calculation means for calculating a feature value representing a feature for each spatial frequency band of the reference block based on the orthogonal data; An image processing apparatus comprising: a correction calculation unit that calculates a correction value based on orthogonal data of a predetermined spatial frequency band; and an attribute determination unit that determines an attribute of the attention area based on the characteristic value and the correction value. . 4. The image processing apparatus according to claim 1, wherein the attribute is one of a halftone area, a character area, and a continuous tone area. 5. The image processing apparatus according to claim 1, wherein the feature calculation unit calculates the feature value excluding the orthogonal data having the lowest spatial frequency band. 6. The image processing apparatus according to claim 1, wherein the image is an image read from a document. 7. The image processing apparatus according to claim 6, wherein the reference selection unit sequentially selects the reference block by shifting one pixel at a time in the order of main scanning from the main scanning direction and in the sub-scanning direction. 8. The method according to claim 1, wherein the orthogonal transformation unit is configured to, when a pixel constituting the image is included only in a partial area in the reference block, determine a value of a pixel in an area other than the partial area in the reference block. Is filled with a value representing a white image.
The image processing apparatus according to any one of claims 1 to 3. 9. The image processing apparatus according to claim 1, further comprising an image processing selection unit that selects an image processing for the attention area based on a determination result of the attribute determination unit. 10. The image processing apparatus according to claim 9, wherein said image processing is compression processing. 11. The feature value of the reference block and the feature value of the other block having the attention area in the image when the area corresponding to the attention area of the other block is outside the image. The image processing apparatus according to claim 2, wherein: 12. An integrating means for integrating the characteristic value and the correction value of the reference block and the characteristic value and the correction value of another block, respectively, wherein the attribute judging means is configured to perform the operation based on the integration result of the integrating means. The image processing apparatus according to claim 3, wherein an attribute of the attention area is determined. 13. The other block, wherein the attention area is around the attention area of the reference block.
The image processing apparatus according to any one of the preceding claims. 14. A procedure for selecting a reference block including a plurality of pixels constituting an image, and selecting a region of interest including one or more pixels in the reference block, and determining a value of a pixel in the reference block. A procedure for obtaining orthogonal data that is a result of the orthogonal transformation; a procedure for calculating a feature value representing a feature for each spatial frequency band of the reference block based on the orthogonal data; and a feature value of the reference block and other blocks. An image processing method, comprising: a step of integrating a characteristic value for each spatial frequency band; and a step of determining an attribute of the attention area based on a result of the integration. 15. A procedure for causing a computer to select a reference block including a plurality of pixels constituting an image and to select a region of interest including one or more pixels in the reference block; Obtaining orthogonal data as a result of orthogonally transforming the values of the above, a step of calculating a characteristic value representing a feature of each of the spatial frequency bands of the reference block based on the orthogonal data, and a feature value of the reference block and the like. An image processing program for executing a procedure of integrating the characteristic value of each block with respect to each spatial frequency band, and a procedure of determining an attribute of the attention area based on a result of the integration.