JP2629183B2 - Image area identification device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus that reads various originals such as characters, photographs, halftone prints, and performs image processing and editing.
In particular, the present invention relates to an image area identification device that identifies the type of a document.

[Conventional technology]

2. Description of the Related Art In an image processing apparatus such as a digital copying machine or a facsimile, various kinds of originals are read, and image processing, editing, and the like are performed, and then recording is performed.

For example, there are many general document documents in which a binary image region such as a character and a halftone image region such as a photograph or a halftone dot are mixed, such as a catalog. When reading and recording such an original, if the entire original is binarized and recorded, the reproducibility of the characters is good, but the quality of the halftone portion is deteriorated. However, halftones are well reproduced, but there is a problem that characters are deteriorated.

For example, characters are binarized so that the quality is good for each of them, and halftones need to be identified with two different areas in order to perform dither processing.

As a method for identifying this area, Japanese Patent Laid-Open No. 58-33
As shown in Japanese Patent Publication No. 74, the screen is divided into blocks composed of a plurality of pixels, and the difference between the maximum value and the minimum value of the pixel density is obtained in each block. If the difference is larger than a certain threshold, There is a method in which a region is determined as a binary image region, and if smaller than a threshold, a region is determined as a halftone image region. Also, Japanese Patent Laid-Open No.
As shown in Japanese Patent No. 220563, the absolute value of the difference between the focus value, which is the central pixel level of the pixel matrix, and the defocus value, which is the average value of the pixel matrix, is taken. A method of determining an area is also known.

According to these methods, the state of change in density gradient is significantly different between a character and a photograph, so that the two can be relatively easily identified.

[Problems to be solved by the invention]

However, any of these methods has a drawback that characters and halftone dots cannot be distinguished because they focus only on a steep change in density gradient. That is, there is a problem that a halftone dot is erroneously recognized as a character because a sharp change in density gradient exists in the halftone dot similarly to the character.

The present invention has been devised to solve the above-described problems, and has as its object to identify characters and halftone dots with high accuracy.

[Means for solving the problem]

In order to achieve the above object, the present invention scans document information through a window having a size extending over a plurality of image units, and sets a maximum density of a plurality of neighboring image units located before a target image unit with respect to a scanning direction in the guide. Means for calculating the maximum value of the values and the shades of a plurality of neighboring image units located after the image unit of interest; determining a level difference between the maximum values; comparing the level difference with a threshold for area identification; Means for determining the target image unit as a binary image area when the difference is larger than the threshold value.

The scanning is performed in the main scanning direction and the sub-scanning direction, and a window is provided one-dimensionally in each of the directions. If at least one direction is determined to be a binary image area, the target image unit is a binary image area. Region.

Further, the threshold value may be changed according to the density of the window.

[Action]

 First, the principle of identifying image regions will be described.

In the present invention, the original is scanned with a window having a size covering a plurality of images such as a plurality of pixels or a plurality of blocks, and the density change of neighboring pixels before and after the pixel of interest at this time, or a plurality of pixels including the pixel of interest. Focusing on the density change of neighboring blocks before and after the target block
The halftone part and the photographic part are separated. In the following description, a case where the density of a pixel is detected will be described as an example.

FIG. 2 shows an example of the density distribution in the horizontal direction, that is, the main scanning direction of an image including the halftone dot portion 21 and the character portion 22. A, B, C, D, and E in the figure indicate pixels scanned by the scanning window 20, and a pixel C is a target pixel. Further, when processing is performed for each block with a plurality of pixels as one block unit, a maximum value, a total sum, an average value, or the like of the densities of a plurality of pixels included in the block can be adopted as the density of each block. .

When this density distribution is observed through, for example, a 5 × 1 scanning window 20, the following is found.

Concentration change in the halftone dot portion 21 is severe, the level difference d ₁ between the maximum value of the halftone dot pitch P is small.

In the character portion 22, the density changes sharply, and the target pixel C
Level difference d ₂ between the maximum value of the front and rear guide is large.

From this, the maximum value of the density of the neighboring pixels A, B and D, E before and after the pixel of interest C is detected, the difference is obtained, and this difference is compared with the threshold for area identification, thereby obtaining the character portion 22. It can be seen that the dot portion 21 can be identified. Further, since the density change is gradual in the photographic part, it can be distinguished from the character part by the above method. Note that the maximum value here includes the peak value.

Such identification is also performed in the vertical direction, that is, in the sub-scanning direction, and when either the main scanning direction or the sub-scanning direction is determined to be a character portion, the region is identified as a character portion, thereby obliquely identifying the character portion. Almost all character components including directions can be detected.

In addition, a threshold for area identification is set to an image unit in a window, for example,
If the threshold value is changed in accordance with the density of the block, the threshold value changes in accordance with the content of the image, so that identification errors are reduced.

〔Example〕

Hereinafter, the features of the present invention will be specifically described based on embodiments with reference to the drawings.

First, FIG. 3 shows a schematic block diagram of the entire image processing apparatus to which the image area identification device of the present invention is applied.

In the figure, reference numeral 1 denotes an image input device using, for example, a CCD (Charge Coupled Device).
It is read as 8-bit grayscale data at a resolution of m (400 dots / inch). Document information read by the image input device 1 is stored in an image memory 2 of 8 bits per pixel.
Is stored in The image memory 2 has a capacity to store one page of image data in A3 size.
The image data from the image memory 2 is supplied to the image processing device 3, and optimal processing is performed according to the document area. The details of the image processing device 3 will be described later. Reference numeral 4 denotes a control device for controlling signals of the entire device.
The image data on which the predetermined processing has been performed by the image processing device 3 is supplied to the image output device 5 and, for example, about 32 dots / mm
A binary image is recorded at a resolution of (800 dots / inch). In addition, as the image output device 5, an electrophotographic laser beam printer, a thermal printer, an ink jet printer, or the like can be used.

Next, a configuration example of the image processing apparatus 3 will be described with reference to FIG.

The image data read by the image input device 1 (see FIG. 3) is temporarily stored in the image memory 2 for one page. Next, the image data is sequentially read from the image memory 2 and branched into four paths a, b, c, d, and a horizontal density difference detection circuit 6, a vertical density difference detection circuit 7, a filter processing circuit
11 and a binarization circuit 12, respectively.

In the path a, the horizontal density difference detection circuit 6 detects the density difference of the image in the horizontal direction, that is, the image in the main scanning direction.
This is compared with a predetermined threshold value th in the comparator 8, and if the density difference is larger than the threshold value th, 1 is output assuming that a character portion exists, and if it is smaller, 0 is output. Note that the comparator 8 is set so as to output either 1 or 0 when the density difference is equal to the threshold th.

Details of the horizontal density difference detection circuit 6 will be described with reference to FIG.

In the figure, reference numeral 30 denotes a window memory 30 for storing density data for five pixels in the main scanning direction, which corresponds to the 5 × 1 window 20 in the main scanning direction shown in FIG. From this window memory 30, the maximum value max1 = max (A _d , B _d ) of the density of the pixels A and B in front of the pixel of interest C is detected by the maximum value detection circuit 16, and Then, the maximum value max2 = max (D _d , E _d ) of the density of the pixels D and E that follow is detected by the maximum value detection circuit 17.
Note that A _d , B _d , D _d , and E _d indicate the density of the pixels A, B, D, and E. As the density of a block when processing is performed for each block, a maximum value, an average value, a total sum, and the like of the densities of a plurality of pixels included in the block can be adopted.

These maximum values max1 and max2 are supplied to a subtraction circuit 18 to determine the absolute value ABS of the difference, and the comparator 8 calculates the absolute value ABS of the difference as a predetermined level difference, that is, a threshold for area identification. It is compared with th, and if it is larger than this, it is determined as a character. Thereby, the character area can be identified in the horizontal direction.

In this case, the size of the window 20 in the scanning direction is the halftone dot pitch P
Is desirably twice or more. For example, when the halftone density of the halftone printing of the document is 133 lines, the halftone dot pitch P is P = 1.
Since inch / 133 = 190 μm, the size of the window 20 is 1
90 μm × 2 = 380 μm or more. Thus, window 2
By setting the size of 0 to twice or more the halftone dot pitch, even if the halftone dot has a low screen ruling, the distinction between the character and the halftone dot can be reliably realized. This is the same in the vertical direction. When the resolution of the image input device 1 is 16 dots / mm, the sampling interval for reading the original is 1 mm / 16 dots =
62.5 μm. Therefore, as shown in FIG.
The number of pixels may be 7 pixels from 380 / 62.5 = 6.08.

On the other hand, in the path b shown in FIG. 4, the vertical density difference detecting circuit 7 detects the density difference in the vertical direction, that is, the sub-scanning direction, and compares the density difference in the same manner as the horizontal density difference detecting circuit 6. Is compared with the threshold value th in the device 9, and when the density difference is larger than the threshold value th, 1
If smaller, 0 is output.

In the OR circuit 10, when at least one of the outputs of the comparators 8 and 9 is 1, the pixel of interest and the block have a character portion if they are 1, and if both are 0, a character portion exists. It outputs 0 as not performing.

By providing the circuit for detecting the density difference in the vertical direction, that is, also in the sub-scanning direction, almost all character components including oblique directions can be detected.

By performing the above processing for each pixel, a character region can be extracted over the entire image.

The above-described horizontal density difference detection circuit 6, vertical density difference detection circuit 7, comparators 8, 9 and OR circuit 10 constitute an image area identification device 15, and outputs 1 in the case of a character. In the case of halftone, 0 is output.

Next, in the path c, processing is performed on the assumption that the image data is a halftone area.

The image data input to the filter processing circuit 11 is nx
m is smoothed by a two-dimensional FIR (finite impulse response) filter. Note that n is the size in the horizontal direction, and m is the size in the vertical direction. Here, in order to remove the halftone components from 133 lines (133 lp / inch) to 175 lines, n =
m = 5, that is, a 5 × 5 low-pass filter is used. In addition, lp means line pairs (line pairs). The coefficients of the filter used in this embodiment are shown below.

FIG. 5 shows the spatial frequency characteristics. As can be seen from the spatial frequency characteristics, the halftone dot component in the halftone region is completely removed, so that the occurrence of moire at the time of image output can be prevented. Note that the size of the filter is not limited to 5 × 5, but is arbitrary. Further, n ≠ m, that is, a coefficient other than a square may be used depending on the characteristics of the input device or the user's intention.

The path d is a processing which is performed on the assumption that the image data is a binary image area such as a character. The binarization circuit 12 binarizes the image data using a fixed threshold Bth. The binarization here converts the value D to the maximum gray value when the image data value D is greater than or equal to the fixed threshold value Bth, that is, 255 in the case of 8-bit data, and to the minimum gray value 0 when it is small. It is something to convert.

In the selection circuit 13, if the output of the image area identification circuit 15 is 1, it is a character, so the signal on the path d is selected, and if the output is 0, the signal on the path c is selected because it is halftone. .

The dither processing circuit 14 dithers the selectively obtained image data with a threshold matrix shown below, whereby the halftone image is dithered, and the binary image such as a character is already converted into the maximum value and the minimum value. So, simply 2
Valued.

With the above-described configuration, halftone dots and the like in the halftone area are smoothed by the low-pass filter of the filter processing circuit 11 and then subjected to dither processing, so that occurrence of moire is suppressed. In addition, a binary image area such as a character can be simply binarized and output. At this time, since the halftone dot portion is not mistaken for the character portion, it is possible to appropriately reproduce the two regions having different properties.

In the above-described embodiment, the threshold th for area identification is set to a fixed value in advance. However, in this state, an identification error may occur around an isolated point caused by noise in a highlight portion of a halftone dot image. That is, since the maximum density itself becomes small in the highlight portion of the halftone dot image, when noise is mixed, the difference between the maximum densities in the window becomes large and may be erroneously recognized as a character. Therefore, the sixth
In the illustrated embodiment, the threshold value th is adaptively controlled according to the density in the window 20.

That is, a threshold storage ROM 31 in which a plurality of threshold data are stored in advance is provided, and each pixel A, B, C,
The adder 32 adds the densities of D and E, and addresses the threshold storage ROM 31 based on the added output, and changes the threshold according to the sum of the densities in the window. Specifically, the threshold is increased in accordance with the decrease in the density. By adaptively controlling the threshold value th in this manner, for example, the threshold value th becomes high because the density is low in the highlight portion of the halftone image, and even when noise is mixed, the halftone image is erroneously recognized as a character. Will not be done. Therefore, the identification accuracy can be increased even for an image in a poor state. The average of the densities may be used instead of the sum of the densities.

In the above-described embodiment, the density is detected on a pixel basis, but the density may be detected on a block basis.

Further, when the present invention is applied to a color image, red,
A luminance signal may be generated from three types of color signals read by the green and blue filters, and this processing may be applied using the luminance signal as a density signal.

〔The invention's effect〕

As described above, in the present invention, document information is scanned by a window having a size over a plurality of image units, and the maximum value of the density of a plurality of neighboring image units before the image unit of interest and the plurality of pixels after the image unit of interest are scanned. Is determined with respect to the maximum value of the neighboring image unit, and when this difference is larger than the threshold for area identification, 2
It is determined to be a value image area. Since a halftone dot has a large change in density but a small difference between the maximum values, the halftone dot can be reliably discriminated without being erroneously recognized as a character. Therefore, if image processing is performed on each area in accordance with the identification result, an image of each area can be reproduced under optimal conditions. In addition, this identification can be realized by a simple process of finding the maximum value of the density of a plurality of image units before and after the target image unit and comparing the difference between the two maximum values with a threshold for area identification. Configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a density difference detection circuit used in an image area identification apparatus according to an embodiment of the present invention, FIG. 2 is an explanatory view showing the principle of image area identification, and FIG. Block diagram of the entire image processing system,
FIG. 5 is a block diagram showing a configuration example of an image processing apparatus according to the present invention, FIG. 5 is a graph showing a spatial frequency characteristic of a two-dimensional FIR filter, and FIG. 6 is a block diagram showing another example of a density difference detection circuit. . 8: Comparator 15: Image area identification device 18: Subtraction circuit 20: Scan window 30: Window memory

Claims (3)

(57) [Claims] 1. The document information is scanned by a window having a size covering a plurality of image units, and the maximum value of the shading of a plurality of neighboring image units located before the image unit of interest in the scanning direction in the guidance and the image unit of interest. Means for determining the maximum value of the shades of a plurality of neighboring images located after, and determining the level difference between the maximum values, comparing the level difference with a threshold for area identification, and the level difference is larger than the threshold. Means for determining the target image unit as a binary image area. 2. The scanning is performed in a main scanning direction and a sub-scanning direction, and the windows are provided one-dimensionally in each of the directions, and when at least one direction is determined as a binary image area, 2. The image area identifying apparatus according to claim 1, wherein the target image unit is determined to be a binary image area. 3. The apparatus according to claim 1, wherein said threshold value is changed in accordance with a density of an image in said window.