JP3605773B2 - Image area discriminating device - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to a technique for determining the type of an image to perform processing according to the type of the image when performing image processing on image data read from a document or the like in a laser printer, a copying machine, or the like.
[0002]
[Prior art]
Conventionally, in a laser printer, a copying machine, or the like, a character area, a photograph area, and a halftone image area are detected by detecting only the character area from the mixed image data. While a character image is obtained, a photograph / halftone image having a gentle gradation is obtained by performing a smoothing process on a photograph / halftone image area (see Japanese Patent Application Laid-Open No. 4-239269). ).
[0003]
In addition, image data subjected to image processing is temporarily recorded in a memory, and output from the memory for reproduction. However, in order to record image data of a large number of documents in the memory, compression processing is performed. In such a case, the character image may be compressed at a high compression rate with binary data, while the photograph / dot image is of high quality. Therefore, it is necessary to perform compression at a low compression ratio in order to ensure image quality.
[0004]
For this reason, various methods have been conventionally proposed as the image discrimination method (JP-A-2-294888, JP-A-2-295354, JP-A-3-262380, JP-A-3-262380). No., etc.).
[0005]
[Problems to be solved by the invention]
By the way, when performing a scaling process for changing a magnification (hereinafter referred to as a scaling ratio) of a reproduced image (displayed on a CRT or a printed image) with respect to an original image, there is also a method of performing interpolation or thinning processing on read image data. However, there is a method of changing the reading speed of the scanner to omit these image processing. That is, when an image is reproduced in an enlarged manner, the image can be enlarged by reducing the reading speed, increasing the number of read pixels per unit length, and reproducing the image while keeping the pixel interval at the time of reproduction constant. Conversely, at the time of reduced reproduction, the reading speed may be increased to reduce the number of read pixels.
[0006]
However, when the reading speed of the scanner is changed as described above, the sampling interval in the reading direction changes, so that the frequency component and the edge direction of the image apparently change, which may affect the determination of the image. there were.
SUMMARY OF THE INVENTION The present invention has been made in view of such a conventional problem, and has as its object to provide an image discriminating apparatus capable of accurately discriminating an image type without being affected by a reading speed of a scanner. I do.
[0007]
[Means for Solving the Problems]
Therefore, the image determination apparatus according to the present invention, pixels from the image information different kinds of image areas obtained by reading a more original image in at least one direction of the image reading speed variably adjustable free image reading means are mixed In an image region discriminating apparatus including a discriminating unit for discriminating a type of an image region for each, an edge image extracting unit for extracting an edge image having a large number of edge components, and a network for extracting a dot image A point image extracting unit, and when the image reading speed for obtaining a reproduced image having the same size as the original image is set as the reference speed, the image reading speed variably adjusted by the image reading unit is equal to or less than the reference speed. together with at least the first range, adjusted to a direction etc. ho halftone image discrimination criterion the image reading speed of the halftone image extraction unit decreases the halftone images of reading winding direction to facilitate the extraction, the image At least a second range image reading speed is adjusted to the variable is less than the reference speed of the winding means, the edge image extraction unit of nearly as read winding direction of the edge component image reading speed is reduced edge image discrimination criteria It is characterized by comprising a judgment reference adjusting means for adjusting in a direction in which extraction is difficult , and wherein the first range is set within a range of 1/2 to 1 times the reference speed .
[0008]
The image reading means may be included in the image area discriminating apparatus. Alternatively, the image reading means may be provided separately from the image area discriminating apparatus so that image reading speed information in the image reading means can be input to the image area discriminating apparatus. Good.
Further, the second range may be set within a range of 2/3 times or less of the reference speed.
[0009]
Further, the image area determination means, when the image reading speed is greater than the reference speed, etc. ho reading speed of the edge image discrimination criterion of the edge image extraction unit decreases the read winding direction of the edge components to easily extracted You may make it adjust to a direction.
Further, the determination criterion adjusting means may set the halftone image determination criterion of the halftone image extraction unit to be constant in a third range where the image reading speed is lower than the first range.
[0010]
Further, the discriminating means may include a character extracting unit for extracting, as characters, an image obtained by excluding the halftone image extracted by the halftone image extracting unit from the edge image extracted by the edge image extracting unit. Good.
[0011]
[Action]
For example, image reading is performed by performing main scanning in the vertical direction while performing sub-scanning in the horizontal direction by a line scanner as an image reading unit, and adjusting the reading speed in the sub-scanning direction according to the scaling factor for the scaling process. If so, the determination criterion adjusting means adjusts the criterion for determining the image area according to the read speed of the adjusted image.
Here, in the halftone dot image extraction unit, the sampling interval decreases as the reading speed decreases in at least the first range which is set within a range of 1/2 to 1 times the reference speed, and the halftone dot image is apparently displayed. In this range, the dot image discrimination criterion is adjusted in a direction in which the dot image in the reading direction is easily extracted in accordance with the decrease in the reading speed, so that the dot image determination accuracy is good. Is maintained.
Further, in the edge image extraction unit, at least a second range of the reference speed, by adjusting the edge image discrimination criterion in a direction difficult to extract an edge image, the type of determination accuracy of the image area is good Secured .
[0012]
Good results were obtained when the second range in which the discrimination criterion was adjusted in a direction in which edge components were hardly generated was set within a range of 2/3 times or less the reference speed .
When the image reading speed is higher than the reference speed, it is not difficult to determine the halftone image, and the edge image determination criterion is adjusted so that the edge component in the reading direction is more easily extracted as the reading speed decreases. Edge images can be accurately extracted.
[0013]
On the other hand, in the third range where the image reading speed is smaller than the first range, good results were obtained by keeping the halftone image discrimination standard constant .
Also, in the case where an image in which a halftone image is excluded from an edge image is extracted as a character, by adjusting the determination criterion as described above, it is possible to maintain good accuracy in the determination between the halftone image and the character. it can.
[0014]
【Example】
An embodiment of the present invention will be described below with reference to the drawings.
FIG. 2 shows a system configuration of the image area discriminating apparatus according to the present invention.
The scanner 1 as an image reading means is constituted by a line scanner, performs main scanning in the line direction, reads image data at a constant period, and relatively moves the line scanner and the original image in a direction orthogonal to the line direction. In this embodiment, the sub-scanning speed, that is, the reading speed in the sub-scanning direction, is variable according to the magnification (magnification of the reproduced image with respect to the original image) in the sub-scanning direction. Can be adjusted.
[0015]
The dot image discrimination reference value adjustment unit 2 adjusts a discrimination reference value for extracting a dot image, which will be described later, according to the reading speed of the scanner 1 in the sub-scanning direction.
The edge image discrimination reference value adjustment unit 3 also adjusts a discrimination reference value for extracting an edge image, which will be described later, according to the reading speed of the scanner 1 in the sub-scanning direction.
The halftone dot image area extraction unit 4 extracts a halftone dot image area from the image data read by the scanner 1 according to the criterion value adjusted by the halftone image discrimination criterion value adjustment unit 2.
[0016]
The edge image area extraction unit 5 extracts a character area having a large number of edge components and a halftone image area from the image data read by the scanner 1 in accordance with the criterion value adjusted by the edge image determination criterion value adjustment unit 3. The combined edge image area is extracted.
The character area extraction unit 6 extracts a character area by masking an edge image area obtained by combining the character area and the halftone image area with the halftone image area. That is, the character area and the halftone dot image area can be determined separately.
[0017]
The photograph area extracting unit 7 extracts an area other than the character area and the halftone dot image area as a photograph image area. More specifically, a photo region is extracted by masking the entire image region with the edge image region.
Next, the function of each unit will be described with reference to the drawings.
FIG. 3 shows a dot image area extraction routine.
[0018]
Step (S in the figure; hereinafter the same) In step 1, the luminance signal of each pixel is input for each unit image area including the pixel of interest and its surrounding pixels. More specifically, the unit image area is an image block of 5 × 5 pixels centered on the pixel of interest (see FIG. 4). At the same time, a threshold value used for binarization for extracting a halftone dot candidate pixel, which will be described later, according to the reading speed of the sub-scanning method when the luminance signal is read by the scanner, and a halftone dot region from the halftone dot candidate pixel A threshold used for binarization for determination is set.
[0019]
In step 2, each luminance signal value (digital value) is inverted (bit inverted). As a result, the higher the luminance, the lower the level, and the lower the luminance, the higher the level. It should be noted that in the inversion processing, edges having a lower maximum density, such as mesh lines and thin lines, can be extracted more favorably at the time of edge extraction described later than the density conversion processing.
In steps 3A and 3B, right diagonal edge extraction and left diagonal edge extraction are performed, respectively. This is processed by a right diagonal edge extraction filter and a left diagonal edge extraction filter, respectively. That is, in the halftone dot image region, since halftone dots are periodically present at a high density in oblique directions (45 °, 30 °, 70 °, etc.) at equal intervals, the edge frequency component in this direction becomes large. Having. In addition, it has frequency components having an angle of about 90 ° with each other. (Here, as shown in FIG. 5, the direction passing through the first and third quadrants is the oblique right direction, and the direction passing through the second and fourth quadrants is We will call it diagonally left).
[0020]
Therefore, a pair of diagonal edge component extracting means for extracting the left and right diagonal edge components by a filter having a large gain of diagonal frequency components in which halftone dots exist at high density in advance is used as a unit around the target pixel. An edge component in the left-right diagonal direction is extracted for each image area.
For example, the right diagonal edge extraction filter is composed of a filter having a coefficient of 5 × 5 as shown in the figure, and a value obtained by multiplying by a coefficient at a position corresponding to the luminance inversion value of each pixel and adding the result is output as an edge extraction component. Is done. This filter is for extracting edge components for halftone dots in the diagonal right direction of 45 °, but is created so as to be able to extract to some extent in the halftone directions of 30 ° and 70 °. That is, FIG. 6 shows a frequency characteristic of the gain of the right oblique edge extraction filter, FIG. 7 shows the frequency characteristics with contour lines, and FIG. 8 shows a case where the pixel interval is 400 DPI. This shows the peak value of the frequency component of the halftone dot in the 45 ° direction. As shown in the figure, the right diagonal edge extraction filter extracts only the right diagonal direction of the frequency component of the halftone dot. In the diagonal left direction, a frequency characteristic for extracting a frequency component in the diagonal left direction can be created only by exchanging the left and right coefficients of the filter.
[0021]
The output values of the edge components extracted by the right diagonal edge extraction filter and the left diagonal edge extraction filter are binarized by comparing them with threshold values in steps 4A and 4B, respectively. That is, a signal of 1 is output when the output value of the edge component is equal to or greater than the threshold value, and a signal of 0 is output when the output value is less than the reference value. Here, as described in step 1, a value variably adjusted according to the sub-scanning speed of the scanner according to the magnification is used as the threshold. Specifically, the sub-scanning speed when the magnification is 100%, that is, when the magnification is equal, is set as the reference speed, and the speed ratio e with respect to the reference speed is in a range of 1 ≦ e ≦ 2 (magnification = 50-100%). In the range of 1/2 ≦ e <1 (magnification = 100-200%), the threshold value is decreased in accordance with a decrease in the speed ratio, that is, a decrease in the sub-scanning speed, and 1/4 ≦ e. In the range of <1/2 (magnification = 200 to 400%), the maximum value is set to be constant. FIG. 9 shows the amount of adjustment in the direction that makes it easier to determine the halftone image with respect to the magnification (speed ratio e) (the magnitude relationship with the threshold value is reversed. That is, the smaller the threshold value, the more easily the halftone dot is determined). Show the relationship.
[0022]
That is, when the reading speed in the sub-scanning direction is changed, the sampling interval in the sub-scanning direction changes, so that the frequency component of the image and the direction of the oblique edge change apparently. In the case of the left and right diagonal edge filters, an edge component in a diagonal 45 ° direction is set to be extracted.However, if the reading speed is made lower than the reference speed to increase the magnification, the frequency component of the edge is shifted to the lower frequency side. Simultaneously with the shift, the image is stretched in the sub-scanning direction, and the diagonal edge direction changes. As a result, the original 45 ° oblique edge component is less likely to be extracted. Therefore, the reading speed in the sub-scanning direction is slow, and the speed ratio e is within the range of 1/2 ≦ e <1 (magnification = 100 to 200%). It is adjusted so that the edge component in the direction is easily extracted, so that the same halftone image discrimination performance as in the case of the reference speed (actual size) is obtained.
[0023]
If the reading speed in the sub-scanning direction is increased (reduced) to less than 100% faster than the reference speed, the sampling interval in the sub-scanning direction increases, and at that point, it is easy to extract the oblique edge components. However, on the other hand, since the oblique edge direction deviates from the 45 ° direction and it becomes difficult to extract the oblique direction edge component of the halftone dot image, the same as the reference speed threshold value, and especially the oblique direction edge component is extracted. Do not make adjustments that make it difficult to perform.
[0024]
In step 5, the AND of the binarized signal value is calculated. That is, when both signal values are 1, it indicates that the oblique direction edge component near the crossing point in the left and right diagonal direction, that is, the point where the halftone dot exists in the case of the halftone image, is large. The target pixel of the block is a halftone dot candidate pixel.
As described above, the unit image area is scanned in the main direction and the sub direction, and the halftone dot candidate pixels are determined for all the target pixels.
[0025]
Next, in step 6, the number of halftone dot candidate pixels in a window of, for example, 75 × 15 pixels is counted.
In step 7, the count value of the halftone dot candidate pixels in the window is binarized by comparing it with a predetermined threshold value. That is, when the count value is equal to or greater than the threshold value, the pixel of interest is regarded as a halftone image area, the pixel value is set to 1, and when the count value is less than the threshold value, a signal of 0 is output as another image area. Also in the above determination of the halftone dot image area, the halftone dot image area is extracted from the entire image by scanning the window in the main direction and the sub direction to make the determination. This threshold value is also set by adjusting the tendency shown in FIG. 9 according to the reading speed in the sub-scanning direction, similarly to the case of the binarization threshold value when extracting the halftone dot candidate pixels in step 1. I have. However, a configuration in which only one of these two binarization thresholds is adjusted may be employed.
[0026]
As described above, the influence of the change in the reading speed can be avoided as much as possible by extracting the halftone dot image area in the oblique direction edge component and adjusting the determination reference value (each of the thresholds) corresponding to the reading speed adjustment. It is possible to accurately and quickly determine.
In step 8, the halftone dot image area extracted as described above is extended. More specifically, the area extending to the outer side by 4 pixels in the vertical and horizontal directions from the outermost pixel determined as the halftone dot image area is extended as the halftone dot image area.
[0027]
That is, in the absence of the extension, when the halftone image is partially present in the document, the area outside the halftone image area enters the window at the periphery of the area and the number of halftone pixel candidates decreases. As a result, it may be erroneously determined that the area is not a halftone image area, and a boundary with a photographic image area or the like may be erroneously determined. By performing the above expansion, the surrounding area is also expanded and determined as a halftone area. Erroneous determination of the boundary with the other image area can be prevented.
[0028]
Next, a routine for extracting an edge image area will be described with reference to the flowchart of FIG.
In step 11, a luminance signal is converted into a density signal for each pixel in a unit image block including a pixel of interest and peripheral pixels. For example, it is obtained by conversion using a conversion LUT (look-up table) stored in a ROM.
[0029]
In step 12, the density signal is gamma converted. This can also be obtained using a simple broken line conversion table shown in the figure.
By performing the pre-processing including the luminance-density conversion processing and the gamma conversion processing, it is possible to distinguish fine characters.
In step 13, the gamma-converted signal is smoothed by a low-pass filter.
[0030]
In step 14, edge extraction is performed. This is performed by finding the maximum value of the deviation (absolute value) with respect to the value smoothed in the step 3.
By calculating the deviation from the smoothed value in this way, it is possible to prevent a line from being erroneously extracted as an edge and erroneously determined as a character image.
In step 15, the maximum value obtained in step 4 is binarized by comparing it with a threshold value. That is, when the value is equal to or more than the threshold value, a signal having a pixel value of 1 is output assuming that an edge is extracted from the target pixel, and when the value is less than the reference value, a signal having a value of 0 is output as not extracting an edge. Here, prior to binarization, the threshold is adjusted and set according to the reading speed in the sub-scanning direction. In other words, if the reading speed is reduced, it becomes difficult to extract an edge. Therefore, when the speed ratio e is within a range of 1 ≦ e ≦ 2 (variable magnification = 50 to 100%), the direction in which the edge speed is more easily determined as the reading speed is reduced. That is, adjustment is made so as to lower the threshold value. However, when the reading speed is reduced to 200% or more and the speed ratio e is reduced to 以下 or less, it becomes difficult to discriminate the halftone dot image and the character. It is determined that it is more preferable to determine the edge as a photographic area. In the range where the magnification ratio is less than 2/3 and the speed ratio exceeds 150%, the edge is adjusted so that it is difficult to extract edges. FIG. 11 shows the tendency of the relationship between the scaling factor (speed ratio e) and the edge image discrimination reference adjustment amount when extracting an edge image.
[0031]
In step 16, the edge image area is extended upward, downward, leftward, and rightward by two pixel widths around the target pixel.
By expanding the edge image area in this way, the effect of applying an emphasis filter to a character area determined as described later is improved.
FIG. 12 shows a photograph area determination routine.
[0032]
In step 21, it is determined for each pixel whether or not the pixel belongs to the edge image area determined in the edge image area determination routine. If the pixel does not belong to the edge image area, the process advances to step 22 to determine that the pixel is a photograph area. .
As described above, even when the image reading speed is changed according to the magnification, the image classification can be performed with stable accuracy by adjusting the image type determination criterion according to the reading speed.
[0033]
Further, in the present embodiment, the present invention is applied to a case where only the reading speed in the sub-scanning direction is variably adjusted, and the scaling process in the main scanning direction may be performed by processing such as interpolation or thinning of image data. . Further, the present invention, of course, can be implemented by variably adjusting the reading speed in the main scanning direction by adjusting the determination criterion in the same manner.
[0034]
【The invention's effect】
As described above, according to the present invention, the image reading speed of the image is variably adjusted according to the magnification of the image, and the image data read by adjusting the determination criterion according to the reading speed. Even if the reading speed is adjusted, stable determination accuracy can be ensured.
[Brief description of the drawings]
FIG. 1 is a block diagram showing the configuration and functions of the present invention.
FIG. 2 is a diagram showing a system configuration according to an embodiment of the present invention.
FIG. 3 is a flowchart showing a halftone image area determination routine of the embodiment.
FIG. 4 is a diagram showing a state when a halftone dot image area is extracted.
FIG. 5 is a diagram for defining an oblique direction of an oblique edge extraction filter.
FIG. 6 is a perspective view showing a frequency characteristic of a gain of the oblique edge extraction filter.
FIG. 7 is a diagram showing the characteristics of the above filter as contour lines.
FIG. 8 is a diagram showing a peak value of a frequency component of a halftone dot in a 45 ° direction.
FIG. 9 is a diagram showing a relationship between a halftone image discrimination reference adjustment amount and a magnification (speed ratio).
FIG. 10 is a flowchart showing an edge image area extraction routine.
FIG. 11 is a diagram illustrating a relationship between a scaling factor (speed ratio) and an edge image determination reference adjustment amount.
FIG. 12 is a flowchart showing a photograph area extraction routine.
[Explanation of symbols]
REFERENCE SIGNS LIST 1 Scanner 2 Halftone image discrimination reference value adjustment unit 3 Edge image discrimination reference value adjustment unit 4 Halftone image area discrimination unit 5 Edge image area discrimination unit 6 Character area discrimination unit 7 Photo area discrimination unit

Claims (5)

The determining means for determining at least one direction of the image reading speed variably adjustable free image reading means on the type of the image area from the image information for each pixel different kinds of image areas obtained by reading more original images are mixed In the image region determination device configured to include
The discriminating means includes an edge image extracting section that extracts a high edge image of the edge component, and a halftone image extracting unit that extracts a halftone image,
When an image reading speed for obtaining a reproduced image of the same size as the original image is set as a reference speed, the image reading speed variably adjusted by the image reading means is at least in the first range equal to or lower than the reference speed, etc. Ho speed read image halftone images discrimination criterion of the halftone image extraction section decreases with adjusting the direction of halftone images of reading winding direction is easily extracted, the image reading is adjusted to a variable of the image reading means at least a second range of speed is less than the reference speed, reading winding direction of the edge components nearly as edge image determination based on the image reading speed of the edge image extraction unit is reduced is adjusted in a direction difficult to extract discrimination criterion Equipped with adjustment means ,
An image area discriminating apparatus, wherein the first range is set within a range of 1/2 to 1 times the reference speed . An image area discriminating apparatus, wherein the second range is set within a range of 2/3 times or less of the reference speed . When the image reading speed is higher than the reference speed, the determination criterion adjusting means adjusts the edge image determination criterion of the edge image extraction unit in a direction in which an edge component in the reading direction is more easily extracted as the reading speed decreases. 3. The image area discriminating apparatus according to claim 1, wherein: The said reference standard adjustment means makes the halftone image determination reference | standard of the said halftone image extraction part constant in the 3rd range where the image reading speed is smaller than the said 1st range, The 1-3 characterized by the above-mentioned. The image area determination device according to any one of the above.   The method according to claim 1, wherein the determining unit includes a character extracting unit that extracts, as characters, an image obtained by excluding the halftone image extracted by the halftone image extracting unit from the edge image extracted by the edge image extracting unit. Item 5. The image area determination device according to any one of Items 1 to 4.

Images