JP3705414B2 - Binarization threshold determination method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for binarizing image data such as characters and images printed on a document in a scanner or the like, and more specifically, in the case of binarizing image data optically detected by an image sensor. The present invention relates to a method for determining a threshold value to be used.
[0002]
[Prior art]
FIG. 7 is a diagram showing an example of the configuration of a scanner which is this type of image reading apparatus. In FIG. 7, reference numeral 10 denotes an image sensor in which a plurality of reading pixels are arranged in the main scanning direction of the figure, and 11 denotes the image sensor 10. An A / D converter that converts an output signal into a digital signal, 12 is a line buffer that sequentially holds the output of the image sensor 10 as data for each line in the main scanning direction, and 2 is a threshold value calculated from the data of the line buffer 12 A threshold calculation unit 3 is a binarization processing unit that outputs binary data of the pixel of interest from the data of the threshold calculation unit 2 and the line buffer 12, 4 is a memory, 5 is a processing circuit, and 6 is a document.
[0003]
In this type of image reading apparatus, when the read data from the image sensor 10 is quantized into black and white binary values, the read data (IMGD) is compared with a predetermined threshold (Th), and the image at the pixel of interest is compared. Is configured to determine whether the data is black data or white data.
If the threshold value used as a reference for binarization is set to a certain level, for example, when the density of the background of the original 6, that is, the density of a blank portion where no image is formed, an image to be originally read is extracted. The problem that it becomes impossible to occur occurs.
In view of this, there is known a method of satisfactorily extracting an image to be read by making the threshold for binarization constant and determining each time according to the background density (dynamic threshold). As this method, for example, as shown in FIG. 8, in the threshold value calculation unit 2, a peripheral pixel (3 × 3) including the line data before and after the pixel of interest (marked with × in the figure) to be read from the line buffer 12 is the center. (9 pixels on the matrix) is read out, the average value of the obtained data is obtained and this is used as the threshold value (Th), and the threshold value Th calculated in the binarization processing unit 3 and the read data IMGD at the target pixel are obtained. In comparison, it is considered to determine monochrome binary data in the target pixel.
[0004]
[Problems to be solved by the invention]
In the method in which the average value is obtained from the above-mentioned data around the target pixel and this is used as a threshold value, when the density difference between the background density and the extracted image is small, or as shown in FIG. When there is an image with high density such as a thick line near the image, the average value of the data calculated as the threshold value approximates or exceeds the density of the image to be extracted. There was a problem that it was impossible to extract the thin line).
[0005]
In order to solve the above-described problems, the present applicant has calculated a threshold value that follows the background density of a manuscript or the like by removing the image data portion to be extracted, thereby enabling accurate binarization. Proposed the following binarization threshold value determination method in Japanese Patent Application No. 11-172145.
That is, the binarization threshold value determining method related to the earlier application calculates an average density value in the target pixel (N) from the image data in the target pixel (N) to be binarized and its surrounding pixels, and this average density value And the difference between the threshold values of the pixel (N−1) immediately before the target pixel (N) to be binarized is used as a change value, and the follow-up indicating the degree of change in the threshold value of the target pixel (N) based on the change value. By calculating the value and adding the follow-up value and the average density, the threshold value for the pixel of interest (N) is obtained, and the change amount of the threshold value increases due to the presence of the image data (black data). Image data can be extracted satisfactorily by making the change in threshold value follow only the change in the background density.
[0006]
However, in this binarization threshold value determination method, since the change in the threshold value is made to follow only the change in the background density, the background density where there is no image to be extracted and the background density approximated the image to be extracted. At locations where the density changes, this is output as image data, which may be noise. That is, in the binarization threshold value determination method of the previous application, the change value ΔTMn is set to the threshold value Th (N−1) in the immediately preceding pixel (N−1) in order to make the change in the threshold value follow the change in the background density. Based on this change value, the threshold value for the pixel of interest (N) is determined. Therefore, as shown in FIG. 10, the density of the background portion where there is no image to be extracted and the background density approximated to the image to be extracted As shown in A of the figure, the input image data exceeds the threshold value, and the image data is output at the location where the input image data changes to the background. There was a case.
[0007]
The present invention has been made to solve these problems, and it is possible to calculate a threshold value that follows the background density of a document or the like by removing an image data portion to be extracted, thereby performing accurate binarization. In addition, the threshold value is determined so that noise is not generated even in a background portion where there is no image to be extracted and even where the background density changes to a density approximate to the image to be extracted.
[0008]
[Means for Solving the Problems]
In order to solve the above-described problem, in the binarization threshold value determination method of the present invention, an average density value in the target pixel (N) is calculated from the image data in the target pixel (N) to be binarized and its surrounding pixels. The difference between the average density value and the threshold value of the pixel (N−1) immediately before the target pixel (N) to be binarized is obtained as a change value, and the threshold value of the target pixel (N) is determined based on the change value. A follow-up value indicating the degree of change is calculated, the follow-up value and the average density are added, and a predetermined addition value is added to obtain a threshold value for the pixel of interest (N).
Further, the added value is obtained based on the difference between the average density in the pixel data string in the main scanning direction including the pixel of interest (N) to be binarized and the average density in the pixel data string in the main scanning direction immediately before. The sub-scanning density follow-up value indicating the degree of change in average density in the sub-scanning direction is calculated, and the calculated sub-scan density follow-up value is added to a preset fixed addition value.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the binarization threshold value determining method of the present invention will be described in detail with reference to the drawings.
Since the configuration of the image reading apparatus to which the binarization threshold value determining method of the present invention is applied may be the same as the configuration of the conventional apparatus shown in FIG. 7, the configuration equivalent to that of the conventional apparatus is described with the same symbol.
[0010]
FIG. 1 is a diagram showing a configuration of a threshold value calculation unit 2 that performs the threshold value determination method of the present invention. An average value calculation unit 21, a change value calculation unit 22, a follow-up value calculation unit 23, a sum calculation unit 24, a threshold value holding unit 25, and The addition value setting holding unit 26 is configured.
[0011]
Next, the threshold value determination method of the present invention will be described based on FIG. Even when image data is captured by the image reading apparatus of the present invention, pixel data stored as data for each line in the line buffer 12 is sequentially read out and quantized into binary values as in the conventional apparatus described above. The threshold value calculation unit 2 calculates a threshold value Th from the unit pixel (target pixel) and the peripheral pixel image data, and the binarization processing unit 3 compares the calculated threshold value Th with the image data IMGD of the target pixel. Thus, it is configured to create binary data.
[0012]
When calculating the threshold value Th in the threshold value calculation unit 2, first, in the average value calculation unit 21, the image data IMGD (N) of the target pixel (N) and a plurality of peripheral pixels centered on the target pixel (N). Image data is read from the line buffer 12. (STEP1)
Next, an average density value Mn (N) at the target pixel (N) is calculated from the n pieces of image data read out by the following Equation 1. (STEP2)
Mn (N) = Σ (IMGD) / n Expression 1
[0013]
Subsequently, in the change value calculation unit 22, the threshold value Th (N-1) of the immediately preceding pixel (N-1) in the main scanning direction of the pixel of interest (N) and the average density value Mn () of the pixel of interest (N) calculated in STEP2. N) is calculated (Equation 2 below), and this is defined as a change value ΔTMn (N). (STEP3)
ΔTMn (N) = Th (N−1) −MnN (N) Equation 2
[0014]
The calculated change value ΔTMn (N) is multiplied by a tracking coefficient fk, which is an arbitrary coefficient, to calculate a tracking value F (N) (Formula 3 below). (STEP4)
F (N) = ΔTMn (N) × fk Equation 3
[0015]
Up to the above STEP 4 is equivalent to the binarization threshold value determination method in the previous application.
Next, the follow-up value F (N) calculated in STEP 4, the average density value Mn (N) in the target pixel (N) calculated in the average value calculation unit 21, and the addition value stored in advance in the addition value setting holding unit Sk is added to calculate a threshold Th (N) for the pixel of interest (N) (Formula 4 below). (STEP5)
Th (N) = F (N) + Mn (N) + Sk Equation 4
[0016]
The calculated threshold Th (N) is output to the binarization processing unit 3 and used for the binarization processing of the image data in the pixel of interest (N). At the same time, the threshold value Th (N) is stored in the threshold value holding unit 25 of the threshold value calculation unit 2 and used for calculating the threshold value in the subsequent pixel of interest (N + 1).
When the calculated threshold value Th (N) is output to the binarization processing unit 3, it is determined whether or not there is a subsequent pixel of interest, that is, an image to be read in STEP6. If there is a subsequent image, the process returns to STEP1 to determine the threshold value. When the process is performed and there is no subsequent image, the threshold value determination process is terminated.
[0017]
According to this binarization threshold value determination method, when determining the threshold value of the target pixel, the difference between the average density value of the image data of the target pixel and its surrounding pixels and the threshold value of the immediately preceding pixel adjacent to the target pixel is calculated. Since it is calculated as a change value, it is possible to reduce the influence that the change amount of the threshold value increases due to the presence of the image data (black data), and to follow only the change in the density of the background, and to extract the image data. It can be done well.
That is, even in the portion that moves from the background portion to the image region, the density change value is obtained by calculating the difference between the average density value around the pixel of interest and the threshold value in the immediately preceding pixel (that is, the background portion), and further to this change value. By setting and multiplying the tracking coefficient, which is an arbitrary coefficient, to an appropriate value, the amount of change in the threshold can be reduced.
This follow-up coefficient fk is an arbitrary value of 0 <fk ≦ 1, and as shown in FIG. 3, by setting the follow-up coefficient fk to a value close to 1 and increasing the follow-up value F (N), the threshold Th The amount of change in the threshold value Th can be suppressed even if the change in the average density Mn increases due to the appearance of the image data, and the threshold value can be made to follow only the background density of the document. Can do.
Further, by appropriately setting the follow-up coefficient fk, the threshold value can be made to follow the change of the average value density Mn, so that, for example, even in an image data region where fine lines are dense, binarization can be favorably performed. .
That is, when the follow-up coefficient fk is set to a value close to 0, the follow-up value F (N) becomes small, and the calculated threshold value Th (N) becomes a value close to the average value concentration Mn, thereby following the change. It becomes. Accordingly, when there are a large number of white and black data in a document in a narrow range, the tracking coefficient is set to a value close to 1, and vice versa, the image data is set to a value close to 0. The amount of change in the threshold value Th can be set as appropriate.
[0018]
Further, in the binarization threshold value determination method of the present invention, a background portion where there is no image to be extracted by adding or decreasing a dynamic threshold value determined for each pixel by adding a predetermined addition value Sk set in advance. Even if the background density changes to a density approximate to the image to be extracted, unnecessary image data is not output and noise is not generated.
That is, in the case of FIG. 4, in order to remove a portion that should not be extracted, the addition value Sk is set as a negative value (the absolute value is a shown in the figure) and added, and the threshold value is shifted in the “black” direction. To do. Thereby, only the image to be extracted can be read and noise can be removed. Conversely, when the number of pixels in the image portion to be extracted is small, such as when the image is faint, the addition value Sk is set and added as a positive value, and the threshold value is shifted in the “white” direction. Thereby, the number of pixels of the image portion to be extracted can be increased.
This added value Sk is set in advance in accordance with the background density in a range where Th (N) calculated by the above equation 4 satisfies the minimum density value ≦ Th (N) ≦ maximum density value. The operator confirms the image state of the document to be read and the background density, and selects and determines the addition value setting holding unit 26 as appropriate, similarly to the tracking coefficient fk.
[0019]
In the above embodiment, the addition value Sk selected and determined in advance by the operator is added as a constant value over the entire reading range of the document. When the background density changes globally in the sub-scanning direction (original transport direction) of the original with uneven background density throughout the entire area, if the addition value Sk is added as a constant value, Depending on the image quality, an unnecessary portion that is not an image to be extracted is generated as noise, or conversely, the number of pixels in the image portion to be extracted is reduced, and the image is blurred.
In order to solve this problem, in the second embodiment of the present invention, the addition value Sk is changed so as to follow the density change in the sub-scanning direction of the document.
[0020]
FIG. 5 is a diagram showing the configuration of the threshold value calculation unit 2 according to the second embodiment of the present invention. Components equivalent to those in the first embodiment are denoted by the same reference numerals.
In the second embodiment, an average density value calculating unit 27 reads out image data LINEIMGD (N) for one line including the target pixel (N) to be binarized from the line buffer 12 and calculates an average value thereof. Reference numeral 28 denotes an average density value holding unit that stores and holds an average density value in the immediately preceding line in the sub-scanning direction including the target pixel (N) to be binarized, and 29 includes the target pixel (N) to be binarized. This is an addition value calculation unit for calculating the addition value Sk_Line (N) in the line to be processed.
In the second embodiment, the tracking coefficient fk and the fixed addition value Sk that determine the degree of change (following) of the threshold Th (N) with respect to the change in the background density of the document are the same as in the first embodiment. It is assumed that the operator has selected and set in advance.
[0021]
Next, based on FIG. 6, a threshold value determining method in the second embodiment of the present invention will be described. Also in the second embodiment, the processing method for calculating the threshold Th (N) is indicated by the same reference numerals as those in the processing method shown in FIG. .
In the threshold value determining method according to the second embodiment of the present invention, before calculating the threshold value of the target pixel (N) to be binarized, one line (that is, pixel data in the main scanning direction) including the target pixel (N) is calculated. The average density value in the column) is calculated, and the addition value Sk_Line (N) corresponding to the background density change in the sub-scanning direction is calculated based on the calculated average density value and the average density value in the immediately preceding line. Configured.
That is, before the threshold value calculation unit 2 calculates the threshold value Th, the average value calculation unit 27 reads out one line of image data LINEIMGD (N) including the pixel of interest (N) to be binarized from the line buffer 12 ( (Step 11), the average density Line_Av (N) in the reading line is calculated from the n pieces of data according to the following formula 5. (STEP12)
Line_Av (N) = Σ (LINEIMGD (N)) / n ... Formula 5
[0022]
Next, in the addition value calculation unit 29, Line_Av (N) calculated by the average density value calculation unit 27 and the average density Line_Av (N−1) of the immediately preceding line stored and held in the average density value holding unit 28 (read) When the line is the first line (N = 1), that is, in the initial state, Line_Av (N) = Line_Av (N−1) (N = N−1)), the average density sub-line of the one line is obtained. A value indicating the degree of change in the scanning direction is calculated, and a sub-scanning density follow-up value is calculated by multiplying this by a preset follow-up coefficient fk. Further, a fixed addition value Sk is added, and an average in the sub-scan direction is calculated. An added value Sk_Line (N) corresponding to the degree of change in density is calculated by the following equation. (STEP 13)
Sk_Line (N) = {Line_Av (N) −Line_Av (N−1)} × fk + Sk Equation 6
Further, the addition value Sk_Line (N) in the line is calculated, the value of the average density value holding unit 28 is updated to the average density value Line_Av (N) calculated in STEP 12, and this is set as Line_Av (N−1). The operation proceeds to the binarization threshold value determining operation (STEP 1 to STEP 5) of the pixel of interest (N). Since the binarization threshold value determination operation in STEP 1 to STEP 5 is the same as that in the first embodiment described above, description thereof will be omitted. In this second embodiment, Sk_Line (N) calculated by STEP 13 as an added value. Therefore, in STEP 5, the threshold value is calculated by the following equation.
Th (N) = F (N) + Mn (N) + Sk_Line (N) Equation 7
[0023]
In the binarization threshold value determining operation, when the threshold value for the pixel of interest is calculated (STEP 5), the process proceeds to STEP 14 to detect whether or not the binarization threshold value determining operation for one line is completed.
Here, if it is determined that there is a pixel of interest to be read in the one line, the process returns to STEP 1 to perform a binarization threshold value determining operation for the next pixel. If it is determined that there is no target pixel to be read, the process proceeds to STEP 15 to detect whether there is a subsequent line.
If it is determined in STEP 15 that the following line data exists, the process returns to STEP 11, the next one line of data is read from the line buffer 12, and the addition value Sk_Line (N) for the line is calculated and the binarization threshold is determined. Perform the action.
If it is determined in STEP 15 that there is no subsequent line data, the binarization threshold value determining operation is terminated.
[0024]
As described above, in the binarization threshold value determining method according to the second embodiment of the present invention, the addition value is made to follow the change rate of the average density for each reading line, that is, the density change in the sub-scanning direction of the document. Therefore, even if the background density changes globally in the sub-scanning direction, no noise is generated or the number of extracted pixels is not reduced.
In the second embodiment of the present invention described above, an example in which the tracking coefficient fk related to density change tracking in the main scanning direction is used when calculating the sub-scanning density tracking value has been described. The tracking coefficient may be set separately from the tracking coefficient in the main scanning direction.
[0025]
【The invention's effect】
As described above in detail, in the binarization threshold value determination method of the present invention, the average density value in the target pixel (N) is calculated from the image data in the target pixel (N) to be binarized and its surrounding pixels, The difference between the average density value and the threshold value of the pixel (N-1) immediately before the target pixel (N) to be binarized is obtained as a change value, and the change degree of the threshold value in the target pixel (N) based on this change value Since the threshold value for the pixel of interest (N) is obtained by calculating the tracking value indicating the value, adding the tracking value and the average density, and further adding a predetermined additional value, this enables accurate binarization. And a threshold value can be determined so that noise is not generated even in a background portion where there is no image to be extracted and where the background density changes to a density approximate to the image to be extracted. When Achieve the cormorant effect.
Further, the difference between the average density in the pixel data string in the main scanning direction including the pixel of interest (N) to be binarized and the average density in the pixel data string in the main scanning direction immediately before the addition value is obtained, and based on this The sub-scanning density follow-up value indicating the degree of change in the average density in the sub-scan direction is calculated, and the calculated sub-scan density follow-up value is added to the preset fixed addition value. Even if the background density changes globally, there is an effect that no noise is generated or the number of extracted pixels is not reduced.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a configuration of a threshold value calculation unit that performs a threshold value determination method according to the present invention.
FIG. 2 is a flowchart showing a threshold value determining method according to the present invention.
FIG. 3 is a diagram showing a threshold value, an average density value, a tracking value, and a change value in the threshold value determining method of the present invention.
FIG. 4 is a diagram showing threshold values obtained by the threshold value determining method of the present invention.
FIG. 5 is a diagram illustrating a configuration of a threshold value calculation unit that performs the threshold value determination method according to the second embodiment of the present invention;
FIG. 6 is a flowchart showing a threshold value determining method according to the second embodiment of the present invention.
FIG. 7 is a diagram illustrating a configuration of an image reading apparatus.
FIG. 8 is an explanatory diagram showing a conventional threshold value determining method.
FIG. 9 is a diagram showing threshold values obtained by a conventional threshold value determining method.
FIG. 10 is a diagram showing threshold values obtained by the threshold value determination method of the previous application.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Image sensor 11 A / D converter 2 Threshold calculation part 3 Binarization process part 4 Memory 5 Processing circuit 6 Original

Claims (2)

Line data is configured by arranging and storing a plurality of pixel data in the main scanning direction, and further storing image data two-dimensionally by storing a plurality of line data in the sub-scanning direction. In the method for determining a threshold value for binarizing the data of each pixel,
From the data of the target pixel (N) to be binarized and its surrounding pixels, the average density at the target pixel (N) is calculated,
The difference between this average density value and the threshold value of the pixel (N−1) immediately before the pixel of interest (N) to be binarized is determined as a change value,
Based on the calculated change value, a tracking value indicating the degree of change of the threshold value in the pixel of interest (N) is calculated,
A binarization threshold value determining method characterized by adding the follow-up value and the average density value and further adding a predetermined added value to obtain a threshold value for the pixel of interest (N). The added value is obtained based on the difference between the average density in the pixel data string in the main scanning direction including the pixel of interest (N) to be binarized and the average density in the pixel data string in the main scanning direction immediately before. The sub-scanning density follow-up value indicating the degree of change in the average density in the sub-scanning direction is calculated, and the sub-scanning density follow-up value is determined by adding the calculated sub-scanning average follow-up value and a preset fixed addition value. Item 2. The binarization threshold value determining method according to Item 1.

Images