JP3134292B2 - Image processing 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 document information by pre-scanning and records image data by main scanning.

[Conventional technology]

2. Description of the Related Art In an image recording apparatus such as a copying machine or a facsimile, not only originals using ordinary white paper but also originals using various papers such as newspaper, straw half-paper, recycled paper, and colored paper are to be read. Documents using such papers other than ordinary paper have a high background density, so if they are read by a document reading means such as a CCD sensor and the image data is output as it is, the reproduced document will have no background. It will be dirty.

Therefore, for a document having such a background with a constant density, conventionally, for example, a certain area such as near the center of the document is read, and the light quantity adjustment for detecting the average light quantity and adjusting the developing bias is conventionally performed. The processing is performed so that the highlight portion of the background is saturated to reduce the density of the background.

[Problems to be solved by the invention]

However, in the conventional background removal method as described above, since only the average light amount in a certain area is detected, the detection level also changes depending on the image density of the document,
There is a problem that it is not possible to correctly determine whether a document has a dark background or a bright document. In addition, if a process of shifting the conversion curve between input and output to saturate the highlight side is performed as shown in FIG. 17, the density is reduced as a whole.

In addition, it is often performed to copy a manuscript in which a cutout of a newspaper, a magazine, or the like is pasted and edited on a white paper. In such a manuscript, a plurality of background densities are mixed. However, conventionally, it has not been possible to remove the background by adjusting the density only in a specific area of such an attached document. Moreover, when trying to remove a relatively high density background, low-contrast parts such as handwritten characters written on a white background are also erased,
It was impossible to remove the background of high density and at the same time leave the white handwritten characters and the like.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has as its object to set a threshold value for background removal for each region. It is another object of the present invention to detect a density region of each background in a document in which a plurality of background densities are mixed, and to be able to switch a threshold value. Still another object of the present invention is to be able to separate the background density of the bonded portion from the low-contrast image of the white portion.

[Means for solving the problem]

For this purpose, the present invention provides a density detecting means for detecting the density of each pixel of a document, a histogram creating means for counting the number of pixels for each density and creating a histogram of the density distribution, and obtaining a peak of the density distribution from the histogram. A threshold determining means for detecting and determining a threshold value for removing the background of the document is provided, and the histogram creating means creates a histogram of only a low density area where the background density of the document is detected.

A density detecting means for detecting the density of each pixel of the document; a histogram creating means for counting the number of pixels for each density to create a histogram of the density distribution; and detecting a peak of the density distribution from the histogram to remove the background. Threshold determination means for determining a threshold value for the density, a histogram creation means divides the density into a plurality of density areas, counts the number of pixels for each density area, creates a histogram of density distribution, Is characterized in that adjacent density regions are set so as to partially overlap each other.

Further, density detecting means for detecting the density of each pixel of the document, histogram creating means for counting the number of pixels for each density and creating a histogram of the density distribution, and detecting the peak of the density distribution from the histogram to remove the background. Threshold value determining means for determining a threshold value for the original document, and the threshold value determining means uses the histogram of the density distribution created by the histogram creating means to detect the high density in the low density area where the background density of the document is detected. It is characterized in that a peak in which the number of pixels equal to or more than the reference value is counted from the region side is detected, and a threshold value for background removal is determined based on the peak.

[Action]

In the image processing apparatus of the present invention, a density detecting unit that detects the density of each pixel of a document, a histogram creating unit that counts the number of pixels for each density and creates a histogram of the density distribution, By providing a threshold value determination unit that detects a peak and determines a threshold value for background removal, even in a bonded document in which a plurality of background densities are mixed, it is possible to detect each background density region, In recording image data, the background density of the document is detected, the threshold is switched, and the background image data is replaced with white.
The background can be efficiently removed for each area.

〔Example〕

 Hereinafter, embodiments will be described with reference to the drawings.

FIG. 1 is a diagram for explaining an embodiment of the image processing apparatus according to the present invention.

In the present invention, as shown in FIG.
The density detection means 1 reads the density of each pixel as document information at the time of pre-scanning, and the histogram creation means 2 creates a histogram 3 from which the background is removed even when newspapers, colored paper, etc. are stuck together. And the background removal circuit 7 is configured to remove the background from the input image data of the original read at the time of the main scan by using the threshold. That is, as the threshold value 5, a value corresponding to each background is determined for the stitched original. At the time of the main scan, the threshold value is switched while judging from which background area based on the value, and the background removal processing is performed. Do. Therefore,
Even if a newspaper is pasted on a white paper original, a low threshold is used for the white paper,
Even a low-contrast image such as a handwritten character is reproduced without being removed, and is switched to a high threshold value in a newspaper bonding portion, so that a background with a high density is removed.

 First, the background representative value detection circuit will be described.

FIG. 2 is a diagram showing a configuration example of a background representative value detection circuit, and FIG.
FIG. 4 is a diagram for explaining a difference in density distribution between various documents, FIG. 4 is a diagram for explaining the degree of background density dispersion by documents, and FIG. 5 is a diagram for explaining a modification of the histogram creation processing. FIG. 6, FIG. 6 is a view showing an example of setting a density area, FIG. 7 is a view showing a change in a histogram due to a difference in setting of a density area, FIG. 8 is a view showing an example of actual creation of a histogram, FIG. Is a diagram for explaining an example of peak detection,
FIG. 10 is a diagram showing a relationship between a density area and a threshold value to be applied.

In FIG. 2, the comparator 11 has, for example, 8 bits, 256 bits.
A density area is detected by comparing the input image data of the gradation with a reference value of density detection in a low density area of 0 to 138, and a counter 12 counts the number of pixels for each detected density area. -1, 12-2, ..... Therefore, when the input image data of the entire original is processed by the prescan, a histogram of the density distribution relating to the read original can be obtained by the counters 12-1, 12-2,.... The BKG determination circuit 13 includes counters 12-1, 12-
2. A peak of the density distribution is detected from the count value of 2,..., And a plurality of different density regions of the background, such as a white background, a newspaper, a colored paper, etc., having the detected peak as a representative value are determined. The circuit 14 determines a threshold based on each density region determined by the BKG determination circuit 13. Therefore, the density detecting means 1 shown in FIG.
In step 11, the histogram creating means 2 sets the counters 12-1, 12-
2. The threshold determining means 4 is connected to the BKG determination circuit 13 and the threshold
It is configured by a hold determination circuit 14.

As described above, in the case of an original in which a newspaper with a white background is cut out and pasted, the density of the background is considerably different.If a threshold value is set to remove the background of the newspaper, for example, handwritten characters written on a white background paper There is a problem that such a low contrast image is also removed. Therefore, in the case of a stitched original having a plurality of background densities, in order to solve the above-described problem so that each background is skipped and low-contrast characters are not thinned, an individual background level is detected. In addition, it is necessary to accurately switch the thresholds corresponding to the plurality of detected background levels.

For example, in the case of a normal original as shown in FIG. 3 (a), as shown in FIG. 3 (b), a high frequency of the background appears in a low density area which is almost white, and in a high density area which is close to black. The frequency of the image appears. Further, when a plurality of originals having different background densities are stuck together as shown in FIG.
As shown in FIG. 5, a certain frequency appears in each of the low-density areas corresponding to the background density of the pasted original document, in addition to the nearly white low-density area. As described above, the feature of the background portion is that the area occupying the entire document is large and the density is lower than that of the image portion. From this, it can be seen that, in the background removal, as described above, a histogram of the density distribution is created in the low-density region of the document, and a peak representative value is detected by searching for a peak from the histogram.

However, the problem in this case is that if the image data of the entire original is processed, the resolution will be about 32 million pixels (= about 4700 x about 6700) at 400 dpi for A3 size, and the data will be enormous. On the other hand, in the case of uneven background such as newspapers and diazo, the distribution is broader than that of a white background showing a steep peak, and the peak is low. For example, the fourth
FIG. 6A shows the frequency distribution of the density in the case of a white background document, and FIG. 5B shows the frequency distribution of the density in the case of a newspaper document. However, in the case of a document having a low contrast character on a white background, a frequency distribution as shown in FIG. 3C is shown, and the density of the character does not change from the density of the background of the newspaper document.
Therefore, when a newspaper is pasted on a white background,
As shown in (1), the density of the background of the newspaper and the low-contrast characters on the white background overlap.

Therefore, first, in order to reduce the data amount, for example, X /
A Y-division type or a thinning type may be adopted. In the X / Y split type, as shown in FIG. 5 (a), a histogram is created for each line in the main scanning direction (X direction) to determine a peak, and each peak value is calculated in the sub scanning direction (Y direction). The histogram may be collected and a histogram may be created. Alternatively, a histogram may be created by counting a predetermined number of peaks (for example, three) from the histogram in the main scanning direction in the sub-scanning direction. This way,
High detection accuracy can be obtained, and in A3 size,
When the resolution is 400 spi, each line is about 4700 pixels, about 6700
Since there are two lines, the amount of data can be greatly reduced and necessary information can be obtained. The thinning type is shown in FIG.
As shown in (1), since sampling is performed by thinning out at intervals of several pixels to create a histogram, the circuit configuration can be simplified. For example, every 96 pixels in the main scan and 24 in the sub-scan
Create a histogram by sampling each line,
In the main scanning direction, a method of searching for the lowest density pixel during sampling and counting it as background density may be employed. In this way, the background data can be sampled even when the sampling point is an image.

In addition, when pixels are counted over all densities with 8-bit, 256-gradation image data, 256 counters are required. However, the number of counters can be significantly reduced by setting the area of the background density which is a general background to be the highest, for example, the area of up to 138 in newspapers. Further, since the density of the background varies depending on the document and the density at which the peak of the histogram appears is shifted, a density region having a width is set and a histogram in the density region is created. Further, as described above with reference to FIG. 4, since the density of the background varies depending on the document, the width of the density area is changed for each density area corresponding to the document, thereby reducing the number of counters. In addition to this, it is possible to perform highly accurate determination and background removal processing even for variations due to originals.

However, even in the case described above, the appearance of the peak changes depending on the method of dividing the region, and the appearance of the background near the boundary of the region is 2.
There is a case where it is difficult to set a region because it is dispersed in one region and a peak is difficult to appear. In consideration of such a situation, for example, according to the original having uneven background,
It is effective to divide the density data of .about.138 into 10 blocks as shown in FIG. 6 (b) so that the density for forming the histogram has various widths. Then, FIG.
As shown in (c) and the table below, when a peak is searched for, and the peaks are searched for, the next two sides are not selected.

FIG. 7 shows an example in which the width of the area is changed. In contrast to the histogram shown in FIG. 7A, a histogram in which the area has a width is shown in FIG. When the boundaries of the regions are shifted, the histogram shown in FIG. 3C and the half-superimposed histogram are shown in FIG. FIG. 8 shows an example in which an actual histogram is created by the above-mentioned area designation.

In peak detection, when the highest peak is detected, the same process is repeated three times by halving the peak.
Alternatively, the same processing is repeated three times while clearing the peak and both sides to 0 and further halving the next two sides. In this way, the influence of the variation is suppressed, and if the detected peak is equal to or less than the reference value, it is ignored to exclude low-coverage ones and noise-like ones. That is,
Even if the original is not a stitched original or if there are only two types of background levels, peak detection can be performed in accordance with that. In addition, as shown in FIG. 9, a peak search may not be performed, and the background area may be determined from the area on the highest density side exceeding the specified frequency. In this case, a continuous area is defined as one unit, and a plurality of units, for example, A and B shown in FIG.
May be processed as the determination result.

The determination of the density region having the selected peak as a representative value and the threshold value applied to the region is performed as follows. That is, assuming that the density data of a, b, and c shown in FIG. 10 (a) are the selected peaks, the density regions A, B, and C including the lower density side have the respective peaks as representative values. Density areas, and a threshold value TH applied to each of these density areas.
A, THB, and THC take values higher than the concentration range. For example, the threshold THA for the density region A of 0 to 11 is 11, 12 to 57
The threshold value THB for the density region B is set to 67, and the threshold value THC for the density region C of 58 to 138 is set to 148. The threshold determination circuit 14 outputs the density areas A, B, and C and the thresholds THA, THB, and THC.

In order to detect the document size in the prescan, the platen cover needs to have a density higher than the background density of the document, for example, gray. In this case, if the histogram is created by limiting the density area as described above, the peak of the background density is not detected at all in reading the OHP. Therefore, conversely, using this fact, if a specific threshold value can be set in advance for the background removal for OHP, the specific peak is determined on the condition that all the peaks of the histogram are below the reference value and no peak is detected. By using the threshold value of, the background can be removed in the case of OHP.

Among the above, an example of processing when the X / Y split type shown in FIG. 5A is adopted will be described.

FIG. 11 is a diagram for explaining the processing by the BKG determination circuit 13 and the threshold determination circuit 14 shown in FIG.

In this process, when the prescan is started, first, a histogram x is created from the input image data for one line, and a maximum frequency area (peak area) thereof is searched (step (2)).

Next, it is checked whether or not the count number in the maximum frequency area of the histogram x is equal to or more than a reference value (MINX; for example, 50). If it is not less than the reference value, the counter of the maximum frequency area of the histogram y is incremented by 1, and if it is less than the reference value, the process jumps to the step.
Then, the count value of the maximum frequency area of the histogram x is
The same process is repeatedly performed until the maximum frequency area of the histogram x is searched three times by returning to the step ス テ ッ プ, and returning to the step (step 〜).

The process moves to the next line, returns to the steps until the last line is reached, and repeats the same processing (step).

The above is the process of creating the histogram y for the entire original.

When the histogram y is created, the first maximum frequency area (No. 1 area) is searched, and it is checked whether the count number is equal to or more than a predetermined value (line number / 16). If the value is equal to or more than the predetermined value, the maximum frequency area of the histogram y (No. 1ar
ea), the counts on both sides thereof are set to 0, the counts on the two neighbors are halved, and if not more than a predetermined value, all maximum frequency areas (No. 1 area, No. 2 area, No. 3 area) Is invalidated and the threshold value is fixed at THMAX (for example, 143). Similarly, a second maximum frequency area (No. 2 area) and a third maximum frequency area (No. 3)
area), and check whether each count number is equal to or more than a predetermined value, and determine each maximum frequency area (No.1area, No.2area,
No.3area) valid / invalid. In this case, even when the count number of the second maximum frequency area (No. 2area) is a predetermined value or more, the count number and the count numbers on both sides are set to 0,
The count number of the next two is halved (step ~).

And each maximum frequency area (No.1area, No.2area, No.3
area) is rearranged from the high-density area to areaA, areaB, and areaC, and as described in FIG. 10, the effective density area is extended from the high-density side to the low-density side to set the applicable density area (step). ~).

Next, the background removal performed when recording image data in the main scan will be described.

FIG. 12 is a diagram showing a configuration example of a background removal circuit, FIG. 13 is a diagram for explaining an algorithm of threshold switching, and FIG. 14 is a diagram for explaining smoothing processing.

In FIG. 12, the FIFO 21 converts input image data into, for example,
57 pixels and the FIFO 34 delay three pixels, and the comparison circuits 22 and 23 compare the value of the first area with the value of the input image data and detect the pixels in the first area. The counter 24 counts up at the output of the comparison circuit 22 and counts down at the output of the comparison circuit 23. Accordingly, the count value is determined by the number of pixels falling into the first area out of the 60 pixels delayed by the FIFO 21. Become. Similarly, the comparison circuits 25 and 26
Compares the value of the input image data with the second area, the counter 27 counts the number of pixels included in the second area out of the 60 pixels, and the comparison circuits 28 and 29 output the third Is compared with the value of the input image data, and the counter 30 counts the number of pixels included in the third area out of the 60 pixels. Here, the first area A, the second area B, and the third area C are, for example, the first threshold THA, the second threshold THB, and the third threshold THC determined by the description of FIG. This is an area set correspondingly. That is, the above-described circuit checks the image data up to 60 pixels ahead of the pixel output from the FIFO 34 in order to determine the switching of the threshold value. The FIFO 34, the comparators 35 and 36, and the counter 37 are circuits for checking ahead of three pixels. The comparators 35 and 36 compare the threshold value of the applicable background area with the pixels other than the background area (black). Is detected, and the counter 37 counts whether or not there is data outside the background (black) within three pixels. When there is data outside the background (black) within three pixels, switching from low density to high density is prohibited, and switching of the threshold value due to the edge portion of the image is prevented.

The threshold determination circuit 33 determines a threshold to be applied by the BKG removal circuit 32 from the values of the counters 24, 27, 30, and 37, and outputs the threshold to the background removal circuit 32. The BKG removal circuit 32 compares the threshold value determined by the threshold value determination circuit 33 with the input image data, and replaces image data having a value smaller than the threshold value with 0. That is, image data having a value smaller than the threshold value is determined to be the background, and is replaced with white image data to be output image data.

A prerequisite for background removal is to reliably follow the density change of the background part and recognize the character part so that information is not lost. Therefore, to achieve this,
It must be possible to distinguish the background and the text. In the case of a character portion, since the pixel at the subsequent pixel always returns to the background level, it is necessary to use this information to remove only the background portion.

Since the background is the minimum density of the original and has a large area over the entire surface of the original, even in the case of an original in which a plurality of background densities are mixed, the threshold of each area and the image data are compared to determine the background density. Area can be determined.
Therefore, the threshold value may be switched based on this determination.

For example, as shown in FIG. 13, the peak detection corresponds to the white background and the background of the pasted newspaper, and the background areas A, B, C,
When the threshold values THA and THB are applied, the background density is lower than the threshold value THA in a white background area, and the background density is lower than the threshold value THB in the newspaper area, but is higher than the threshold value THA. Therefore, if it is smaller than either threshold,
The smaller threshold value is set as the applied value, and when the value becomes larger than the smaller threshold value, switching should be performed so that the upper threshold value is set as the applied value. However, since the character portion also becomes larger than the threshold value at that time, it is necessary to determine whether the character portion is the original portion of the background having the threshold value above the character portion. Therefore, this determination can be made by detecting the density of pixels in a certain range. That is, as shown in FIG. 13, when image data exceeding the applicable threshold at that time appears, it is to check whether there is a pixel in the area determined by the applied threshold for pixels up to 60 pixels ahead, for example. 60 pixel resolution
If the width is 16 dots / mm, the width is slightly less than 4 mm. Therefore, if the original is a character document, the background is always present within this width. Therefore, in this case, the threshold should not be switched,
If there is no pixel in the area up to that point even when looking at the width of 60 pixels, the threshold is switched to the upper threshold. In FIG. 12, the counters 24, 27, and 30 detect this pixel, and the threshold value applied by the Threshold determination circuit 33 is switched based on this value.

However, the above switching is performed only when switching from the low density threshold to the high density threshold, and conversely, when switching from the high density threshold to the low density threshold, the switching is unconditionally performed. This is to prevent unnecessary use of the high-density threshold value in view of the fact that a character having a density lower than the threshold value on the high-density side may exist on the low-density side. In addition, when switching from the low-density threshold to the high-density threshold, switching is not performed even when there is image data larger than the background area within three pixels from the pixel. This is to prevent switching of the threshold value due to the edge portion of the image,
The image data of the three-pixel ahead check circuit composed of FIFO31 is
It is taken into the Threshold determination circuit 33 and executed.

Further, even after the threshold is switched from the low density to the high density, the threshold may be changed according to the number of pixels of the low density area included up to 60 pixels ahead. In this case, for example, if the coefficient is k and the number of pixels included in the low density area is N, Is corrected by multiplying the coefficient k by a threshold value. However, if the threshold value is equal to or smaller than the threshold value of the low density area, the threshold value is set to the low density threshold. By doing so, it is possible to reduce a sense of discomfort during inappropriate control.

Further, when the threshold value is switched from the low density to the high density, if the background is removed by comparison with a new threshold, a feeling of incompatibility occurs with the background that cannot be skipped. That is, when the white background is changed to a newspaper or a blue diagram, an edge portion or noise that cannot be completely skipped appears. Therefore, if the image data below the threshold is not simply removed as the background, but the threshold is gradually skipped, discomfort can be reduced. FIG. 14 shows the relationship between input image data and output image data when this processing is performed. This process does not output the input image data exceeding the threshold TH as it is, as is apparent from the figure, but performs a process for smoothing the input image data 1.5 times the threshold TH. , Input image data is threshold
If it is less than TH, the output image data is set to 0, and the threshold TH is set to 1.5
If the value is twice or more, the output image data is used as it is, and the difference between the input image data and the threshold value is tripled at the middle to obtain output image data. That is, if input image data IN and output image data OUT are set, OUT = (IN−TH) × 3.

FIG. 15 is a diagram for explaining an example of processing by the threshold determination circuit 33 and the BKG removal circuit 32.

At the start of the main scan, the Threshold determination circuit 33 and the BKG removal circuit 32 first set the area A in a register Area Before that stores the area of the previous pixel (step).

Next, an area “Area Now” to which the pixel to be processed belongs is obtained, and it is checked whether the area is another area or an area of the previous pixel (step to).

And for other areas, Area B is added to Area Now
efore is set, and in the case of the area of the previous pixel, the suppression processing is performed as it is. Area Before is Area Now
If smaller than the threshold th indicated by Area Now, the threshold TH
(Step,-).

In other cases, the Area Befor in the preceding 60 pixels
The number of data belonging to the area indicated by e is checked from the value of the counter, and "threshold th indicated by Area Now-count value x 8"
Is set as the threshold value TH. In other words, the threshold value is corrected by this processing (step).

Compare the threshold TH with the threshold th indicated by Area Before,
The suppression process is performed with the larger one as the threshold value TH (step to).

Set Area Now to Area Before, return to the step until one line ends, repeat the same process,
When one line is completed, the process returns to the steps until the next line is also completed by one screen, and the same processing is repeated (steps to).

In the suppression processing, the processing target image data D is
It is checked whether it is greater than 1.5 times, and the processing target image data D
Is larger, the processing target image data D is output as it is, but if the processing target image data D is smaller than the threshold value TH, the processing target image data D is output as 0, and the processing target image data D becomes 1.5 times as large as the threshold value TH. If it is between
The processing target image data D is replaced with a value of (D-TH) × 3 and output.

FIG. 16 is a diagram showing an example of a system configuration of an image recording apparatus to which the present invention is applied.

In FIG. 16, an IIT 41 has, for example, a full-color CCD sensor and converts an original image into a video signal and outputs the video signal. An A / D converter 42 converts an analog signal output from the CCD sensor of the IIT 41 into an analog signal. It is converted into a digital signal of bits. Shading correction circuit 43
Is for correcting unevenness in the light amount of the light source, unevenness in the sensitivity of the CCD sensor, and the like. The density adjustment circuit (AE) 46 performs the background removal processing according to the present invention. The density adjustment circuit (AE) 46 creates a density distribution histogram by pre-scanning, determines a background removal threshold, and performs main scanning on the background density area. The threshold value is switched accordingly, and background removal is performed. The digital filter 47 detects and enhances edges of characters and lines, and removes halftone dots from a halftone image. The TRC 48 controls tone reproducibility, and the IOT 49 outputs a signal processed in the preceding stage as a copy. In addition, when adding the function of the enlargement / reduction processing and the editing processing, an editing control circuit for that is inserted and connected after the TRC 48, for example.

〔The invention's effect〕

As is apparent from the above description, according to the present invention, a density detecting unit that detects the density of each pixel of a document, and a histogram creating unit that counts the number of pixels for each density and creates a histogram of a density distribution And a threshold determining means for detecting a peak of a density distribution from the histogram and determining a threshold value for background removal. Therefore, even in a bonded document in which a plurality of background densities are mixed, each background density region is In the recording of image data, the background density of the document is detected, the threshold is switched, the background image data is replaced with white, and the background can be efficiently removed for each area.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining an embodiment of an image processing apparatus according to the present invention, FIG. 2 is a diagram showing a configuration example of a background representative value detection circuit, and FIG. FIG. 4 is a diagram for explaining the degree of background density variance due to the original, FIG. 5 is a diagram for explaining a modification of the histogram creation processing, and FIG. 6 is a setting of the density area. FIG. 7 is a diagram illustrating an example, FIG. 7 is a diagram illustrating a change in a histogram due to a difference in setting of a density region, FIG. 8 is a diagram illustrating an example of actual creation of a histogram, and FIG. 9 is a diagram illustrating an example of peak detection. FIG. 10 is a diagram showing a relationship between a density region and a threshold to be applied, FIG. 11 is a diagram for explaining processing by a BKG determination circuit and a threshold determination circuit shown in FIG. 2, and FIG. 12 is a background removal circuit. FIG. 13 is a diagram showing an example of the configuration of FIG. FIG. 14, FIG. 14 is a diagram for explaining the smoothing process, FIG. 15 is a diagram for explaining an example of the process by the threshold determination circuit 33 and the BKG removal circuit 32, and FIG. 16 is an image recording to which the present invention is applied. FIG. 17 is a diagram showing an example of the system configuration of the apparatus, and FIG. 17 is a diagram for explaining a conventional example of background removal processing. 1... Density detecting means 2... Histogram creating means 3
... histogram, 4 ... threshold value determination means, 5 ... threshold value,
6 ... Background representative value determination circuit, 7 ... Background removal circuit.

Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04N 1/40-1/409 H04N 1/46 H04N 1/60

Claims (4)

(57) [Claims] 1. A density detecting means for detecting the density of each pixel of a document, a histogram creating means for counting the number of pixels for each density to create a histogram of a density distribution, and detecting a peak of the density distribution from the histogram. An image processing apparatus comprising: a threshold value determining unit that determines a threshold value for background removal of a document; and a histogram creation unit creates a histogram of only a low-density region in which the background density of the document is detected. . 2. A density detecting means for detecting the density of each pixel of a document, a histogram creating means for counting the number of pixels for each density to create a histogram of the density distribution, and detecting a peak of the density distribution from the histogram. Threshold value determining means for determining a threshold value for removing background color, and the histogram creating means creates a histogram of the density distribution by dividing the density into a plurality of density areas and counting the number of pixels for each of the density areas. An image processing apparatus characterized by the above-mentioned. 3. An image processing apparatus according to claim 2, wherein said plurality of density areas are set by partially overlapping adjacent density areas. 4. A density detecting means for detecting the density of each pixel of a document; a histogram generating means for counting the number of pixels for each density to generate a density distribution histogram; and detecting a density distribution peak from the histogram. Threshold value determining means for determining a threshold value for background removal, and the threshold value determining means uses a density distribution histogram created by the histogram creating means to detect a background density of a document in a low density area. An image processing apparatus comprising: detecting a peak counting the number of pixels equal to or greater than a reference value from the high-density region; and determining a threshold value for background removal based on the peak.