JPH09212642A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To secure satisfactory images in every area, e.g. an image having its clearly emphasized edges in a character area, etc., even with an original where the character, photo and halftone dot areas coexist by performing the θcorrection processing and the digital filter processing to the input image data. SOLUTION: The image of a reading original 1 is formed on an image sensor 3 via a lens system 2 whose focus, etc., are previously adjusted. The image signals fetched by the sensor 3 are amplified at an image amplifier part 4 and then converted into the digital signals via an A/D conversion part 5. These digital signals are inputted to an image processor 17. In such a constitution, the digital filter processing, i.e., the edge emphasis processing and the smoothing processing or the θ correction processing, i.e., the correction processing of density conversion into deep and light colors are carried out in the character, photo and halftone dot areas, respectively. Thus, the picture quality is improved for the input images.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention can perform image processing suitable for obtaining a good output image on a document in which a character area, a photograph area and a halftone dot area are mixed in an input image. Regarding the device.

[0002]

2. Description of the Related Art In a conventional image processing apparatus, various corrections are generally performed on image data input by a scanner or the like. For example, various kinds of deterioration such as deterioration of the optical system when reading a document, generation of aliasing distortion due to frequency limitation due to digital sampling, deterioration of the reproduction system (developing system) during output, and generation of moire due to the effect of halftone dot processing. Occurs. Therefore, compensating for such image deterioration and faithfully reproducing the definition of the image has become a major problem in the field of image processing apparatuses.

Therefore, in recent years, it has become important to perform so-called digital filter processing such as edge enhancement processing and smoothing processing on image data digitized by an image processing apparatus. Here, the digital filter processing is a correction technique for digital multi-valued image data that is performed in order to reproduce a good definition against various kinds of deterioration as described above, and is roughly classified into high frequency (high Edge enhancement processing for enhancing the edge portion of the image by enhancing the frequency component, and vice versa, by suppressing the high frequency (high frequency) component of the image data, the edge portion of the image is gently expressed. , Smoothing processing for the purpose of removing noise.

Further, when outputting multivalued image data (gradation data) in conventional image processing, so-called γ correction, which corrects the density of the output image, is also effective for compensating for image deterioration due to the characteristics of the output device. It is generally used as a means. However, for example, in the case of processing the whole with a deep color, in the photographic area, the gradation is crushed and the reproducibility deteriorates.
In the character area, there were problems such as lines being blurred and unclear. Further, in the case of the edge emphasis processing, the character area has a clear and good image, but in the photographic area, there is a problem that the reproducibility of the halftone is deteriorated. Note that the photographic area here means an image area in which the density continuously changes like a photograph, not to mention a photograph,
It also includes paintings, design drawings, and computer graphics.

Therefore, in the case where a manuscript includes a character area and a photograph area, a technique (for example, Japanese Patent Laid-Open No. 3-88569) of using different values for the .gamma.-correction coefficient for each area has been proposed. There is.

[0006]

However, in the conventional area discrimination, there is a problem in so-called halftone image processing. The halftone image referred to here is a pseudo binary representation by replacing the gradation data of a halftone image such as a photograph with the area of a dot. That is,
The halftone dot image is essentially a binary image of white or black, and is therefore determined to be a character region in the conventional region determination,
The γ correction for the character area is performed, and as it is, the aliasing distortion of the high frequency component and the moiré caused by the interference of the fundamental frequency component such as dither which is the re-halftone dot processing used at the time of output and the halftone dot frequency component. There was a case where a striped pattern called "is generated, and the image was significantly deteriorated. The present invention has been made in order to respond to the improvement in the image quality of output images, which is becoming more and more demanding. Specifically, even in a document in which a character area, a photographic area, and a halftone dot area are mixed, the character area has a sharp edge. Is intended to obtain an image in which the halftone is faithfully reproduced in the photographic area and a good image free of moire in the halftone dot area. Further, the gamma correction processing is not performed on the character area and the halftone dot area in the read document, but only on the photographic area, so that the gamma correction processing unit can be downsized and the cost can be reduced.

[0007]

An image processing apparatus according to the present invention is an image data discriminating means for discriminating a character area, a photograph area and a halftone dot area from input image data, and a photograph by the image data discriminating means. A .gamma.-correction processing section having a selection means for selecting either the area data conversion section or the area data conversion section other than the photograph area, and the digital filter coefficient for the character area and the halftone dot area by the signal of the image data discrimination means. The digital filter coefficient selecting means for selecting the digital filter coefficient and the digital filter coefficient for the photographic area, and the digital filter processing section having a calculating section for performing an arithmetic operation based on the digital filter coefficient selected by the selecting means are provided. The γ correction process and the digital filter process are performed on the image data. Further, according to the present invention, in the γ correction processing unit, when the input image data is determined to be a photographic region by the image data determining unit, the photographic region data conversion unit is determined to be a region other than the photographic region. If
It has a selecting means for selecting the input image data.

With these configurations, for example, in each of the character area, the halftone dot area, and the photographic area, the edge emphasis processing and the smoothing processing as the digital filter processing, or the density conversion and the color correction for correcting the darkness as the γ correction processing are performed. Even a seemingly contradictory correction process such as a density conversion for correcting a thin image can be separately processed for each area, which can greatly contribute to the improvement of the image quality of an output image. Further, even in the halftone dot area, smoothing corresponding to the frequency of the halftone dot can be selectively performed, and a good output image without moire can be obtained. Also, for the character area and the halftone dot area, the γ correction processing is omitted by separately performing digital filter processing,
It can also contribute to size reduction and cost reduction of the device.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. Embodiment 1 of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an image processing device 17 and its peripheral devices according to a first embodiment of the present invention. In FIG. 1, 1 is a read document, 2 is a focus, a magnification, and a read position are adjusted in advance. The lens system functions to form an image of the read original 1 on the image sensor 3. Further, the image signal (analog signal) taken in by the image sensor 3 is amplified by the image amplification section 4, converted into a digital signal in the subsequent analog / digital conversion section 5, and inputted to the image processing device 17. .

The input image data Qin input to the image processing device 17 is then sent to the image data discriminating means 6 for discriminating a character area, a halftone dot area and a photographic area and a γ correction processing section 7, respectively.

Next, in the operation of the image data discriminating means 6, the operation for discriminating between the photograph area and the area other than the photograph area will be described with a concrete example. First, in order to explain the operation of the image data discrimination means 6, FIG. 2A shows a sample in which a unit of 5 × 5 pixel groups is used as an example of image data. The 5 × 5 pixel group as shown here is input image data Qi that is normally input as data for each line.
It can be obtained by providing the image data discriminating means 6 with a line memory for recording data delayed by 5 lines with respect to n. Here, for example, when m is the density of the pixel of interest, a to y are pixel densities at respective positions around the pixel of interest, and the 5 × 5 pixel group shown in A of FIG. This is one unit for determining whether the area is a text area or a photo area.

Next, FIG. 3 is a flow chart showing an example of a method of discriminating a photograph region and a region other than the photograph region in the image data discriminating means 6. First, in the image data discriminating means 6, a unit of pixel density data to be discriminated from the data of the input image data Qin, here, a to y.
Is recorded. Here, a to y are usually multi-valued data, and in the following description, it is assumed that 256 gradations of 0 to 255 are used, 0 represents black, and 255 represents white. The procedure will be described below with reference to FIG. (Procedure 1) First, one unit of data (here, 5 × 5 of a to y) fetched from the input image data for image data determination is acquired.
Data), find the maximum and minimum values. Next, a value obtained by subtracting the minimum value from the maximum value is obtained, and this value is compared with a preset threshold value TH1 (for example, 150). If it is large, it is determined as an area other than the photograph area, and if it is smaller, the next ( Proceed to step 2). (Procedure 2) Next, the maximum value is compared with a preset threshold value TH2 (for example, 50), and if it is smaller, it is determined to be an area other than the photographic area, and if it is larger, the procedure proceeds to the next (Procedure 3). (Procedure 3) Next, the minimum value is compared with a preset threshold value TH3 (for example, 200), and if it is large, it is determined to be an area other than the photograph area, and if it is smaller, it is determined to be the photograph area.

Here, TH1 is a parameter for separating the photographic area and the area other than the photographic area, and the area other than the photographic area is used because the density change of the area other than the photographic area is larger than that of the photographic area. This is for distinguishing from the photographic area. That is, as the TH1, a value slightly larger than the density change value in a general photographic area is selected. Also,
TH2 and TH3 are areas other than the photographic area when it is determined that all pixels in one unit (here, a 5 × 5 pixel area of a to y) for image data determination are almost white or black. In other cases, the image area is determined to be the photographic area. That is, in (Procedure 2), TH2
Is selected to be relatively close to black, that is, close to 0. By comparison, all data in one unit for judgment is TH
When it is smaller than 2 (black), even if the density change of all data in one unit is small, the one unit is judged to be a region other than the photographic region. In (Procedure 3), TH
3 is relatively close to white, that is, a value close to 255 is selected. By comparison, when all the data in one unit for judgment is larger than TH3 (white), the density change of all the data in one unit is selected. At least one unit is determined to be a region other than the photographic region.

Next, as an operation of the image data discriminating means 6, an operation for discriminating a halftone dot region and a character region from a region other than the photographic region discriminated as the photographic region in FIG. 3 will be described with reference to the flowchart of FIG. . First, one unit determined to be an area other than the photographic area by (procedure 1), (procedure 2), and (procedure 3) is followed by all the pixel data (in this case, one pixel) in one unit by (procedure 4) of FIG. 5x from a to y
5), simple binarization is performed depending on whether TH4 is larger or smaller than TH4. Here, for TH4, the median value of the normal hierarchy (here, 127, for example) is selected.
Next, in (procedure 5), the number of change points of the binarized data in (procedure 4) is counted in the X-axis direction. For example, although B in FIG. 2 is a diagram simulating a character area, in FIG. 2B, the number of change points in the X-axis direction is two per column, which is counted as 2 × 5 = 10 as a whole. Furthermore, the number of change points in the Y-axis direction is zero. Next, in (procedure 6), the threshold value THX which is the number of change points in the X-axis direction set in advance and the count value 10 which is the number of change points in the X-axis direction are compared, and the number of change points which is set in the Y-axis direction is set in advance. Is compared with a count value 0, which is the number of change points in the Y-axis direction, and if the count value is greater than the threshold value in both the X-axis direction and the Y-axis direction, it is determined as a halftone dot area. Judgment, and other than that is judged to be a character area.

Here, since the threshold value THX in the X-axis direction and the threshold value THY in the Y-axis direction are generally circular dots,
It is preferable to select the same value in the X-axis direction and the Y-axis direction. For example, in the case of B of FIG. 2, THX = THY =
If it is set to 15, the area B in FIG. 2 is determined to be a character area by (procedure 6). 2C is a diagram simulating a halftone dot area. In FIG. 2C, the count value and threshold value in the X-axis direction and the count value and threshold value in the Y-axis direction are the same as in FIG. 2B. Comparing
The count value is 20 in both the axial direction and the Y-axis direction, and the area C in FIG. 2 is determined to be a halftone dot area by (procedure 6) as compared with the threshold value 15 set above. In this way, the character area and the halftone dot area can be distinguished by utilizing the symmetry of the halftone dot area with respect to the X and Y axes.
In the case of a document or the like having different scales in the X-axis direction and the Y-axis direction, it can be determined by separately selecting a threshold value according to the ratio between the X-axis direction and the Y-axis direction. Also, when a character region is included in one unit in an oblique direction, it becomes symmetrical with respect to the X-axis direction and the Y-axis direction similarly to the halftone dot region, but in the case of the character region, the number of change points is small compared to the halftone dot region, For example, in the case of THX = THY = 15, the number of change points is usually 15 or less, so that the character area can be determined.

Next, the operation of the γ correction processing section 7 will be described. FIG. 5 shows an edge enhancement process as an example of γ correction performed on a region other than the photographic region, and FIG. 6 shows a process for correcting the pixel density as a whole as an example of γ correction performed on the photographic region. Indicates. In FIG. 5, a γ correction coefficient showing the pixel density data after γ correction is shown as a table for each pixel density data in the input image data Qin before correction. Here, for example, gradation 4 is used as pixel density data.
Data is gradation 1 and data of gradation 127 is gradation 1
27 and the data of the gradation 252 are corrected to be the gradation 254. Also, the graph shows the relationship between before and after correction for all pixel density data represented by gradations from 0 to 255. The pixel density data before correction (that is, individual pixel density data in Qin is shown on the X-axis). ), And the pixel density data after correction is shown on the Y axis. In the example of FIG. 5, the halftone data is corrected to be closer to black or white, so that the character becomes a clear image with sharp edges, that is, image processing suitable for a region other than the photographic region. Becomes

Also in FIG. 6, similarly to FIG. 5, the table shows the γ correction coefficient indicating the pixel density data after the γ correction with respect to the individual pixel density data in the input image data Qin before the correction. Here, an example will be described in which the gradation of the halftone data is maintained as it is and the gradation is raised as a whole. In FIG. 6, for example, as the pixel density data, gradation 4 data is corrected to gradation 19, gradation 30 data is corrected to gradation 127, and gradation 252 data is corrected to gradation 254. The graph shows the relationship between before and after correction for all pixel density data represented by gradations from 0 to 255, where the X-axis is the uncorrected pixel density data (that is, individual pixel density data in Qin). And the Y axis shows the corrected pixel density data (that is, the output of the γ correction unit). That is, specifically, for example, when 0 is black and 255 is white, the density of the entire color can be made lighter, so that, for example, the error of the image sensor 3 or the output device is corrected and the halftone is maintained. The image processing is suitable for the photographed area. Further, in the above case, the correction in which the graph is convex is explained, but if the correction in which the graph is convex downward is necessary, the gradation of the halftone data is maintained as it is. At the same time, it is possible to increase the density of the entire color, which makes it possible, for example, to correct an error of the image sensor 3 or the output device and perform image processing suitable for a photographic area in which halftone is maintained. As described above, the γ correction processing unit 7 of the present invention can arbitrarily correct the color density according to the error of the recording system and the characteristics of the apparatus.

Next, FIG. 7 shows the configuration of the γ correction processing unit 7.
In FIG. 7, the γ correction coefficient for the area other than the photographic area shown in FIG.
Further, the .gamma. Correction coefficient for the photographic area shown in FIG. 6 is stored in the photographic area data conversion unit 9, respectively. Further, the selector 11 receives a signal based on the result of the discrimination by the image data discriminating means 6, and the data conversion unit 9 for the photographic region or the data conversion unit 10 for the region other than the photographic region.
Select either.

The photo area data conversion unit 9 selected by the selector 11 or the area data conversion unit 10 other than the photo area receives the signal of the input image data Qin as an address and outputs the stored data as a signal. For example, in the photo area data conversion unit 9 of FIG. 6, when the input image data Qin is data of gradation 4, the data of gradation 19 stored in the address 4 of the photo area data conversion unit 9 is converted. Further, when the input image data Qin has the gradation 30, the data of the gradation 127 stored in the address 30 of the photo area data converter 9 is outputted.
As a result, the image data of the photograph area portion is lightly corrected in the density of the entire color while maintaining the halftone, thereby preventing the gradation of the photograph area portion from being collapsed, for example.

Next, the operation of the digital filter processing section 8 will be described in detail with reference to the target pixel and peripheral pixels shown in A of FIG. First, there is an edge emphasis process as a digital filter process for emphasizing an edge part of a character original or a line drawing to obtain a sharp and clear image. As an example in which four pixels around a pixel of interest are used as reference pixels, It is shown in (1) and will be described below. X = m + α × {(m−1) + (m−h) + (m−n) + (m−r)} ... Equation (1) In the above equation (1), X is the correction data to be obtained. , If formula (1)
When the correction result by becomes a negative value, set X = 0,
If X exceeds the maximum value that m can take (255 in this case), X is set to the maximum value. That is, in the first embodiment, the gradation is set to 0 to 255 after the correction. Further, m is the pixel density of interest, l, h, n, r
Represents the pixel density of the position adjacent to the pixel of interest, and α represents the digital filtering coefficient. Here, the digital filtering coefficient α is a positive value (for example, +0.
25), X is an image in which the density difference with surrounding pixels is emphasized compared with the original pixel data m of interest, that is, a corrected image in which edges are emphasized, and the value of the digital filtering coefficient α is negative. If it is a value (for example, -0.25), X
Is an image with less difference from the surrounding pixel than the original pixel data m of interest, that is, a smoothed image.

Further, in the above formula (1), an example in which only one pixel adjacent to the pixel of interest is referred to for correction is shown. However, in the edge emphasis processing and smoothing processing, the range of pixels to be referred to for correction -Various patterns can be corrected by changing the position and the value of the digital filtering coefficient α.

Next, FIG. 8 shows a concrete configuration of the digital filter processing section 8. In FIG. 8, reference numerals 14, 15, 16 denote memories for processing coefficients for halftone dot area, character area, and photograph area, respectively. The digital filter processing coefficient .alpha.
It is stored as a halftone dot area coefficient and a photograph area coefficient, and is selected by the selector 12 in response to a signal from the image data discriminating means 6 and output to the arithmetic unit 13. Arithmetic unit 1
Reference numeral 3 receives the digital filter processing coefficient from the selector 12 and the image data from the γ correction processing unit 7, performs the calculation of the equation (1), and outputs the result as the output image data Qou.
Output as t.

With the above construction, when the photographic region is subjected to the γ correction process for the photographic region and then subjected to the digital filter process for the photographic region, for example, the edge enhancement process, the halftone is maintained. A sharp and clear image can be obtained, and when a smoothing process is performed as a digital filter process, a natural image in which the gradation of a photograph is retained can be obtained. In addition, in the character area, after edge enhancement processing is performed by γ correction for areas other than the photograph area, further edge enhancement processing is performed as digital filter processing for the character area, resulting in good edge enhancement. You can get a nice image. In the halftone dot area, after performing edge enhancement processing by γ correction for areas other than the photograph area, smoothing processing, especially the frequency area part of the halftone dot, is performed as digital filter processing for the halftone dot area. As a result, a good image without moire can be obtained.

Further, in the configuration of the present invention, the combination of the γ correction processing and the digital filter processing is not limited to the above example, and various combinations can be considered depending on the desired image. For example, a good character area image can be obtained by correcting the overall color density to be light or dark as the γ correction processing of the area other than the photograph area and then performing the edge emphasis processing on the character area as the digital filter processing. .

Embodiment 2 FIG. 9 shows the γ correction processing unit 7 according to the second embodiment. The difference between FIG. 9 and FIG. 7 is that in FIG. 9, there is no area data conversion unit 10 other than the photographic area, and therefore the area determined to be an area other than the photographic area is data that has not been subjected to γ correction processing, that is, In this case, the input image data Qin is directly selected by the selector 11 and sent to the next digital filter processing unit 8. In the case of this configuration, in the area other than the photograph area, it is preferable that the digital filter processing unit 8 in the next stage performs edge enhancement processing on the character area and smoothing processing on the halftone area. As a result, as in the first embodiment, it is possible to obtain a sharp and clear image while maintaining the halftone, and also to obtain a good image in which sharp edges are emphasized in the character region. It is possible to obtain a good image without moire even in the halftone dot area. Image processing apparatus 1 configured in this way
In the γ correction processing unit 7, since the area data conversion unit 10 for areas other than the photograph area can be omitted, the circuit scale can be omitted, and downsizing and cost reduction can be achieved.

[Brief description of drawings]

FIG. 1 is a block diagram showing an image processing device and peripheral devices of the present invention.

FIG. 2 is a diagram showing a pixel density of interest and pixel densities around it.

FIG. 3 is a view showing a flow chart for discriminating a photograph region and a region other than the photograph region in the image data discriminating means of the present invention.

FIG. 4 is a view showing a flowchart for discriminating a character region and a halftone dot region in the image data discriminating means of the present invention.

FIG. 5 is a diagram showing an example of γ correction processing performed on a character area portion.

FIG. 6 is a diagram showing an example of γ correction processing performed on a photograph area portion.

FIG. 7 is a block diagram showing a γ correction processing unit of the present invention.

FIG. 8 is a block diagram showing a digital filter processing unit of the present invention.

FIG. 9 is a block diagram showing a γ correction processing unit according to the second embodiment of the present invention.

[Explanation of symbols]

1 reading original, 2 lens system, 3 image sensor, 4 image signal amplifying section, 5 analog / digital converting section, 6 image data discriminating means, 7γ correction processing section, 8 digital filter processing section, 9 photo area data converting section, Data converters for areas other than 10 photo areas, 11, 12 selectors, 13 arithmetic units, 14, 15, 16 memories, 17 image processing devices.

Claims (3)

[Claims] 1. An image data discriminating unit for discriminating a character region, a photo region and a halftone dot region from input image data, and a photo region data converting unit and a region data other than the photo region by the image data discriminating unit. .Gamma.-correction processing section having a selection means for selecting either of the sections, and a digital filter coefficient for the character area, a digital filter coefficient for the halftone dot area, and a digital filter coefficient for the photographic area are selected by the signal of the image data discrimination means. Digital filter coefficient selecting means and a digital filter processing section having a calculating section for performing arithmetic processing based on the digital filter coefficient selected by the selecting means, and the γ correction processing and the digital filter processing for the input image data. An image processing device for applying. 2. An image data discriminating means for discriminating a character region, a photograph region and a halftone dot region from the input image data, first discriminates a region other than the photograph region and the photograph region from the input image data, and thereafter, The image processing apparatus according to claim 1, wherein a halftone dot area and a character area are discriminated. 3. An image data discriminating means for discriminating a character area, a photograph area, and a halftone dot area from the input image data, and either the photograph area data converting section or the input image data by the image data discriminating means. .Gamma.-correction processing section having selection means for selecting, and digital filter coefficient selection for selecting digital filter coefficient for character area, digital filter coefficient for halftone dot area and digital filter coefficient for photographic area by the signal of the image data discrimination means. An image processing apparatus comprising: a means and a digital filter processing section having a calculation section that performs calculation processing based on the digital filter coefficient selected by the selecting means.