JPH05292309A - Binarizing processing unit - Google Patents

Abstract

PURPOSE:To separate a binarizing area and an intermediate area regardless of a line width of a line pattern by taking presence of an edge and a density gradient history into account so as to discriminate whether or not each picture element is to be binarized. CONSTITUTION:Gradient history information obtained by a gradient discrimination means 112 is stored in a history storage means 113. A discrimination means 114 discriminates whether or not a relevant picture element is a binary picture based on the history information and the result of detection by an edge detection means 111, and gives an instruction to a binarizing processing means 101 accordingly. The storage means 113 stores the gradient history information from the density increase gradient discrimination to the density decrease gradient discrimination. Moreover, while the said history information is stored and the detection means 111 detects an edge, the discrimination means 114 discriminates the said picture element to be the binary picture. In this case, the part between both edges of a linear pattern forming a character or a line drawing is discriminated to be the binary picture based on the storage history information. Thus, a binary picture area and an intermediate area are separated regardless of the line width of the line pattern.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a binarization processing device for binarizing multivalued image data read from an original by an image reading device such as an image scanner.

In recent years, in order to meet the demand for applications such as a document file system and a desktop publishing system, not only a document represented by only black and white binary values such as characters and line drawings but also halftones such as photographs are included. There is a need for an image scanner having the function of reading a document.
In response to this, an image scanner configured to read a document as a multi-valued image has been realized.

On the other hand, an output device such as a display device or a printer device for outputting an image often supports only a black and white binary image, so that a multi-valued image is converted into a binary image inside an image scanner. There is a need.

Here, as a method of binarizing a multi-valued image, a method of simply binarizing based on the magnitude relation with a fixed threshold value and a method of generating a pseudo halftone image using dither are used. There is. The binarization method using a fixed threshold can clearly reproduce a document represented by only black and white binary, but when applied to an image including halftone, the image quality is significantly deteriorated. On the other hand, the method using dither can reproduce halftones such as photographs in a pseudo manner, but if it is applied to a multi-valued image corresponding to a document represented by only black and white binary, it will produce text and line drawings. It may be difficult to read due to the net.

Therefore, when a binary image such as a character or a line drawing and a halftone image such as a photograph are mixed in one original, the multivalued image corresponding to this original is interleaved with the area of the binary image. It is necessary to divide into a toned image area and perform binarization processing using a binarization method suitable for each.

[0006]

2. Description of the Related Art Conventionally, a multi-valued image read by an image scanner is displayed on a display device or the like to ask the operator's judgment, and a binary image area and a halftone image are displayed in accordance with the operator's instruction. A method of dividing a region and binarization has been used.

Further, in the outline portion of a character or a line drawing, it is used that the image data sharply changes from a white level corresponding to white to a black level corresponding to black, or conversely from a black level to a white level. There is also a method of detecting an edge part from a multi-valued image and binarizing only this edge part using a fixed threshold.

For example, the difference between the image data of each pixel and the image data of the surrounding pixels is obtained, and when the difference value is equal to or more than a predetermined threshold value, it is determined that there is an edge at the position of the corresponding pixel, The image data of this pixel may be binarized using the fixed threshold value described above, and the image data of other pixels may be binarized using the dither method or the like.

[0009]

By the way, as described above, in the method in which the operator specifies the division of the binary image area and the halftone image area for each document, the operator's burden is large, In particular, when inputting a large number of documents at one time using a document file system or the like, it is impossible to deal with them at all.

On the other hand, since the method of applying the fixed threshold value only to the edge portion can be automated, it can be applied to the input of a large amount of documents all at once, and the small characters and the line width of 2 pixels or less can be used. Characters and line drawings with can be reproduced clearly. However, when this method is applied to a large character or a character and a line drawing having a line width of 3 pixels or more, the dither method is applied to the image data of each pixel in the central portion of the linear pattern forming the character or the line drawing. I will end up. For this reason, the image data in the central portion of the linear pattern may be set to the white level, which should be the black level, resulting in a so-called "middle-out" state, and the image quality is significantly reduced. Deteriorates. In addition, when the above-mentioned hollow portion occurs, the character recognition processing becomes impossible.

It is an object of the present invention to provide a binarization processing device which separates a binary image area and a halftone image area with high accuracy and binarizes them by a method suitable for each.

[0012]

FIG. 1 is a block diagram showing the principle of the present invention. According to the present invention, in accordance with input of image data representing the densities of pixels arranged two-dimensionally, each pixel is treated as a binary image or a halftone image, and binarization processing of the image data is performed. In a binarization processing device including means 101, edge detection means 111 for detecting an edge at each pixel position in response to input of image data of each pixel and its surrounding pixels, and in response to input of image data A history relating to the density gradient in the pixel arrangement direction according to the inclination determination means 112 for determining the inclination of the density gradient at each pixel position in at least one pixel arrangement direction and the determination result input by the inclination determination means 112. History holding means 113 for holding information and edge detecting means 111
It is determined whether or not each pixel is a binary image based on the detection result by the history storage unit 113 and the history information stored in the history storage unit 113, and the pixel corresponding to the binarization processing unit 101 is binarized as a binary image. Judgment means 11 for instructing whether or not the digitization process should be performed
4 is provided.

[0013]

According to the present invention, the history holding means 113 holds the history information regarding the gradient obtained by the inclination judging means 112, and the judging means 114 is determined from this history information and the detection result obtained by the edge detecting means 111. Determines whether or not the corresponding pixel is a binary image, and sends an instruction to the binarization processing means 101 according to the determination result.
Here, the history holding unit 113 should hold history information indicating that there is a density gradient from the slope determination unit 112 determining the slope at which the density increases until the slope at which the density decreases. Good. Further, if the determination unit 114 determines that the corresponding pixel is a binary image while the history information indicating that there is a density gradient is held and when the edge detection unit 111 detects an edge. Good.

In this case, since the above-mentioned history information is held in the part between the edges on both sides of the linear pattern forming the character or line drawing, each pixel in this part is together with each pixel in the edge part. It is determined to be a binary image. Therefore, regardless of the line width of the linear pattern, it is possible to separate the binary image region and the halftone image region with high accuracy and perform binarization using a method suitable for each.

[0015]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 2 shows the configuration of an embodiment of an image scanner to which the binarization processing device of the present invention is applied.

In FIG. 2, the image reading unit 210 of the image scanner is a CCD line image sensor (CCD).
211 reads the original 201 by dividing it into a plurality of strip-shaped regions (hereinafter referred to as lines) in the main scanning direction corresponding to the arrangement direction of the elements, and the analog-digital conversion circuit (A / D) 212 The analog signal from the CCD line image sensor 211 is converted into digital data and sent as image data of each pixel forming each line described above.

Further, in FIG. 2, the image data from the image reading section 210 has three latches 221 connected in series.
It is input to the shift memory 222a for one line via a, 221b and 221c. This shift memory 22
The output of 2a has three latches 221d connected in series,
It is input to the shift memory 222b for one line via 221e and 221f, and further, this shift memory 222b
The output of is three latches 221g, 22 connected in series.
It is input to the first stage of 1h and 221i.

That is, nine latches 221a to 221
In i, image data corresponding to a region of 3 × 3 pixels in the original 201 is held, and these image data are stored in the edge detecting unit 111 and the inclination determining unit 112 as image data of the pixel of interest and surrounding pixels. It is configured to be sent out. However, in FIG.
221i is indicated by the symbol "L".

In FIG. 2, the edge detecting means 111 is
It is composed of an adder circuit 231 and a ROM 232.
The adder circuit 231 includes latches 221b, 221d and 221.
The image data held in each of f and 221h is added, and the addition result and the image data held in the latch 221e are input to the ROM 232. Here, the image data held in the above-mentioned latch 221e is the image data of the pixel of interest, and the latch 22
The image data held in 1b, 221d, 221f, and 221h is image data of 4 pixels around the target pixel. Therefore, for each address of the ROM 232, the result obtained by subtracting the image data of the pixel of interest indicated by the lower m bits of the address input from the sum of the image data of the 4 pixels indicated by the upper n bits of the address input is equal to or greater than the predetermined threshold value. By holding whether or not, the detection result indicating whether or not there is an edge at the position of the pixel of interest is obtained as the output of the ROM 232.

Further, referring to FIG. 2, the inclination judging means 112 is provided.
The two determination circuits 240m and 240s are configured to determine the inclination in the main scanning direction and the inclination in the sub scanning direction, respectively.

FIG. 3 shows a detailed configuration of the discrimination circuits 240m and 240s. The determination circuit 240m includes two subtraction circuits 2
41 _am , 241 _bm and 6 AND gates 242 _am- 2
It is formed from 42 _fm . In addition, in FIG.
221b, 221d, 221e, 221 shown in FIG.
The outputs of f and 221h are shown as "latch b", "latch d", "latch e", "latch f", and "latch h".

The above-mentioned subtractor 241 _am includes a latch 221.
The output of the latch 221e is subtracted from the output of f, and a signal indicating whether the subtraction result is a positive value, zero, or a negative value is output through the three output terminals O +, Oz, and O-. It is configured to output. Similarly, the subtractor 241 _bm subtracts the output of the latch 221d from the output of the latch 221e,
A signal indicating the subtraction result is output to three output terminals O +, Oz,
It is configured to output via O-.

Here, these subtracters 241 _am and 241
_bm is the output terminal O +, Oz, when the subtraction result is positive.
Logic "1", logic "1", logic "0" from O-
, And outputs logical "1", logical "0", logical "1" when they are zero, and logical "0", logical "1", logical "1" when they are negative. It may be configured to.

The output terminals O + and O- of the subtracter 241 _am are AND gates 242 _am and 242, respectively.
_It is connected to one of the input terminals of _bm , and the other of the input terminals of these AND gates 242 _am and 242 _bm is connected to the output terminal Oz of the subtractor 241 _am . Similarly, the output terminals O +, Oz, O- of the subtractor 241 _bm are connected to the input terminals of the AND gates 242 _cm , 242 _dm , respectively.

These AND gates 242_am, 242_cm
Output of AND gate 242_emHas been entered in
And gate 242_emBut two subtractors 241_am, 2
41 _bmTo determine whether the subtraction results from both are positive values.
It is configured to set. Here, the above subtraction result is
The fact that both are positive values means that in the vicinity of the pixel of interest,
Indicates that the image data increases monotonically in the main scanning direction.
And AND gate 242_emOutput of image data
The inclination in the main scanning direction indicates whether or not the inclination is positive.

Similarly, the AND gate 242._bm, 242_dm
Output of AND gate 242_fmHas been entered in
And gate 242_fmBut two subtractors 241_am, 2
41 _bmDetermine whether the subtraction results from both are negative values.
This AND gate 242 has a fixed configuration._fmof
Due to output, the inclination of the image data in the main scanning direction is negative
It is indicated whether or not.

The discriminator circuit 240s has two subtractors 241 _as and 241 _as in the discriminator circuit 240m described above.
_bs and six AND gates 242 _{as to} 242 _fs, and the inclination in the sub-scanning direction is obtained based on the image data of three pixels continuous in the sub-scanning direction which are input from the latches 221b, 221e, and 221h. It is configured to determine.

The outputs of the AND gates 242 _em and 242 _fm of the discriminating circuit 240m are set-reset type flip-flops (hereinafter simply referred to as "flip-flops") 2 as the discrimination result by the discriminating circuit 240m.
It is input to the set terminal S and the reset terminal R of 43.

Therefore, this flip-flop 243
Is set to a logic "1" according to the determination result that it is a positive inclination, and then holds the set information until it is reset according to the input of the determination result that it is a negative inclination. It is composed. That is, the flip-flop 243 holds history information indicating that the change of the image data in the main scanning direction has a positive inclination, and fulfills a part of the function of the history holding unit 113.

On the other hand, the history information in the sub-scanning direction is, for example,
The history holding circuit 245 including the line memory 244 for one line is provided, and the determination result for each pixel of one line is stored in the line memory 244 based on the outputs of the AND gates 242 _es and 242 _fs of the determination circuit 240s described above. Just keep it. The line memory 244 may be configured to output the determination result of the corresponding pixel of the previous line held in the line memory 244 according to the input of the new determination result.

The detection result of the above-mentioned edge detection means 111, the flip-flop 243 and the line memory 24.
The output of 4 is the OR gate 2 corresponding to the determination means 114.
It is input to each of the three input terminals 51.

Therefore, the output of the OR gate 251 is logical "1" when at least one of the detection result by the edge detecting means 111 and the history information of the inclination in the main scanning direction and the sub scanning direction is input. It becomes "1", and becomes logical "0" only when these inputs are all logical "0".

That is, when the logical "1" is held as the history information in the main scanning direction or the sub-scanning direction in addition to the edge at the position of the pixel of interest, the logical "1" is output as the output of the OR gate 251. Is obtained. If this is a determination result that the image is a binary image, in addition to the case where there is an edge of the linear pattern at the position of the pixel of interest, the pixel of interest is also detected when the pixel of interest is in the middle of the edges. It is possible to obtain a determination result indicating that is a binary image.

As described above, the history information of the gradient of the density gradient in the main scanning direction and the sub scanning direction is held, and it is determined whether or not the pixel of interest is a binary image while considering the history information. As a result, regardless of the line width of the linear pattern, the linear pattern portion can be separated as a binary image without omission. This makes it possible to separate the binary image area and the halftone image area with high accuracy even in an image obtained by reading a document including characters having thick strokes, and it is possible to perform binarization suitable for each. It is possible to perform binarization processing using the method.

For example, as shown in FIG. 2, a binarization processing means 101 is formed by including a binary image processing unit 261, a halftone processing unit 262 and a selector 263, and the binary processing unit 101 is formed according to the output of the OR gate 251 described above. The selector 263 may be configured to select the image data processing result by the binary image processing unit 261 and the image data processing result by the halftone processing unit 262 and output the result as the binarization processing result.

The above-described binary image processing unit 261 may be configured to binarize the image data based on the magnitude relation between the fixed threshold value and the input image data. Further, the halftone processing unit 262 may be configured to generate a pseudo halftone image from the input image data using dither.

In this case, all the pixels included in the portion of the linear pattern having a thick stroke are judged to be binary images by the above-mentioned judging means 114, and accordingly, the selector 263 judges that they are binary images. The output of the value image processing unit 261 is selected. Therefore, regardless of the line width of the linear pattern forming the character or line drawing, the corresponding portion can be surely binarized by the method using the fixed threshold.

As a result, it is possible to prevent the occurrence of "blank" in the central portion of the stroke of a large character, and to significantly improve the image quality after binarization. The image read by the image scanner can be directly used as a document file. It is possible to meet the desire to use it in a system or desktop publishing system. In addition, by preventing the "blank" of large characters and characters with thick line width, the recognition rate of characters in the document file system, etc. can be improved, and users can save time and manage documents efficiently. it can.

[0039]

As described above, according to the present invention, it is determined whether or not each pixel is treated as a binary image by taking into consideration the presence or absence of an edge and the history regarding the density gradient. Therefore, the line width of a linear pattern is determined. Regardless, the area of the binary image and the area of the halftone image can be separated with high accuracy. Therefore, it is possible to binarize all the pixels included in the linear pattern portion as a binary image, prevent hollow characters with thick strokes, and improve the image quality of the binarized image. Can be planned.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a configuration diagram of an embodiment of an image scanner to which the binarization processing device of the present invention is applied.

FIG. 3 is a detailed configuration diagram of a determination circuit.

[Explanation of symbols]

 101 Binarization Processing Means 111 Edge Detection Means 112 Inclination Discrimination Means 113 History Holding Means 114 Judgment Means 210 Image Reading Units 211 CCD Line Image Sensors (CCD) 212 Analog-Digital Conversion Circuits (A / D) 221 Latches (L) 222 Shift memory 231 Adder circuit 232 ROM 240 Discrimination circuit 241 Subtractor 242 AND gate 243 Set-reset type flip-flop 244 Line memory 245 History holding circuit 251 OR gate 261 Binary image processing unit 262 Halftone processing unit 263 Selector

Claims (1)

[Claims] 1. A binary process for binarizing image data of each pixel as a binary image or a halftone image in response to input of image data representing the density of each pixel arranged two-dimensionally. In a binarization processing device including a conversion processing unit (101), an edge detection unit (111) that detects an edge at a position of each pixel according to input of image data of each pixel and its surrounding pixels, Inclination discrimination means (112) for discriminating the inclination of the density gradient at the position of each pixel in the arrangement direction of at least one pixel according to the input of the image data, and the input of the discrimination result by the inclination discrimination means (112). Accordingly, history holding means (113) for holding history information regarding the density gradient in the pixel arrangement direction, detection results by the edge detection means (111) and the history holding Whether or not each pixel is a binary image is determined based on the history information held in the stage (113), and the pixel corresponding to the binarization processing unit (101) is binarized as a binary image. Determination means (114) for instructing whether or not to be processed
And a binarization processing device.