JP3484831B2 - Binary image multilevel reduction processing device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a binary image multi-value reduction processing apparatus, and can be applied to, for example, reducing and displaying a black-and-white binary image on a low resolution display having a multi-gradation display capability.

[0002]

2. Description of the Related Art Conventionally, for example, a monochrome binary image is reduced and displayed on a low resolution display having a multi-gradation display capability.
2. Description of the Related Art As a value image multi-value reduction processing apparatus, there is one that filters an input binary image to make it multi-valued, and then performs thinning processing to display the reduced image as a multi-valued image. Figure 2 is 2
As a value image multi-value reduction processing apparatus, for example, Japanese Patent Laid-Open No. 2-2
A reduced character display device disclosed in Japanese Patent No. 67592,
In the figure, 1 is a character generation circuit, 2 is a filter circuit, and 3 is a display device.

The binary character image generated by the character generation circuit 1 is filtered by the filter circuit 2 to convert the binary character image into a multi-valued image, and subsequently a desired reduction ratio (decimation ratio). , And the reduced image is output as a multi-valued image on the display device 3. In practice, the filter circuit 2 of such a reduced character display device uses a so-called average value filter for calculating an average value of pixel values in a local area determined according to a thinning rate, an error diffusion filter, or the like.

[0004]

DISCLOSURE OF THE INVENTION Problem to be Solved by the Invention
The filter circuit 2 of the value image multi-value reduction processing apparatus uses an average value filter, which is not very good as a low-pass filter, or an error diffusion filter, which is not a low-pass filter, to make a binary image multi-value. Therefore, the image after multi-value quantization is not always smooth, and the problem that the visibility is locally deteriorated cannot be avoided.

Further, regarding the average value filter, in order to handle the pixel values in the support of the filter with uniform weighting,
It is not possible to multi-value the pixel array within the filter support, and for the same reason, the number of filter outputs is limited to the number of filter support sizes plus 1, so the display grayscale is limited. There was a problem that even if the number was large, it could not be fully utilized.

Further, the multi-value quantization process corresponds to an operation of blurring and smoothing an image. However, after the multi-value reduction processing, blurring also occurs around a thick line segment that is not necessarily blurred. However, there is a problem in that the image looks blurry as a whole. Furthermore, when thinning out the result obtained by applying the filtering process to a thin line segment, in the thinning-out process that thins out pixels that are periodically positioned at intervals that match the thinning rate, the information of the thin line segment is sometimes May be missing
There was a problem that the visibility as a whole was still insufficient.

The present invention has been made in order to solve the above-mentioned problems, and it is possible to significantly improve the visibility when a multi-value reduced image is obtained by thinning after binarizing a binary image. An object is to provide a reduction processing apparatus. further,
It is an object of the present invention to provide a multi-value reduction processing apparatus capable of effectively removing a loss of information of a thin line caused by blurring or thinning occurring in the process of multi-value reduction processing.

[0008]

In order to solve the above problems, a binary image multi-value reduction processing apparatus according to the present invention is a data format conversion means for converting an input binary image into a data format of a multi-valued image. And filter determining means for determining a low-pass filter suitable for band-limiting the multi-valued image according to a desired thinning rate M: N, which is a reduction rate N / M (N and M are positive integers), and data. A multi-valued image obtained by the format conversion means is filtered by the low-pass filter determined by the filter determination means to take out low-frequency signal components, and a multi-valued low-frequency image is generated by giving an intermediate gradation to the multi-valued image. It is provided with a binarization processing means and a thinning processing means for thinning the multi-valued low-frequency image obtained from the multi-valued processing means according to a desired thinning rate M: N to generate a multi-value reduced image.

Further, a data format conversion means for converting an input binary image into a data format of a multivalued image, and a desired thinning rate consisting of an input reduction rate N / M (N and M are positive integers). M: N is a ratio a ^k : 1 (a and k are positive integers), and a thinning ratio M: N is a ratio a ^k : 1 as a result of the thinning ratio judging unit. In this case, a filter determining unit that determines a low-pass filter suitable for band limitation for thinning out a multi-valued image at a thinning rate a: 1, and a multi-valued image that is subjected to filter processing by the low-pass filter determined by the filter determining unit are often used. The binarization process and the thinning-out process of thinning out at the thinning-out ratio a: 1 each time the filtering process is repeated k times respectively, and finally the thinning-out ratio a
^{The present} invention is provided with a recursive iterative filter processing and thinning processing means for outputting a multi-value reduced image thinned out to ^k : 1.

If the thinning-out ratio M: N is not the ratio a ^k : 1 as a result of the thinning-out ratio judging means, first, the recursive iterative filter processing and thinning-out processing means
A multi-valued image is thinned out at a bk ^' ratio of 1 ( ^bk' < ^ak ) (b,
k ′ is a positive integer), and the multi-value reduced image obtained as a result of the thinning is continuously thinned at a thinning rate M: Nb ^{k ′} as a thinning processing, and finally the thinning rate M: N. It outputs a multi-value reduced image thinned out to.

If the thinning-out ratio M: N is not a ratio a ^k : 1 as a result of the thinning-out ratio judging means, first, the recursive iterative filter processing and thinning-out processing means
A multi-valued image is thinned out at a bk ^' ratio of 1 ( ^bk' < ^ak ) (b,
k ′ is a positive integer), and the multi-value reduced image obtained as a result of the thinning is continuously thinned by the nearest neighbor method or the projection method at a thinning rate M: Nb ^{k ′} .
Finally, the multi-value reduced image thinned to the thinning ratio M: N is output.

Further, in the case of thinning-out thin-line determining means for determining a portion for a thin line having a thinning-out rate or less in an input binary image, and thinning-out the result of low-pass filter processing for a portion determined to be a thin line by the thin-line determining means. And adaptive thinning processing means for locally and adaptively changing the thinning method.

Further, the fine line determination means determines whether or not each pixel of the input binary image belongs to a line segment having a width equal to or less than the thinning rate in each of the vertical and horizontal directions, and determines in either direction. If the line width is less than or equal to the thinning rate, the corresponding pixel is determined to be a line segment portion.

Further, when the thinning-out ratio M: N is thinned out, the adaptive thinning-out processing unit leaves only pixels, which are periodically positioned at intervals corresponding to the thinning-out ratio M: N, among pixels which are not determined to be thin line portions. The thinning processing is performed as described above, and as for the portion including the pixel determined as the thin line portion in the thinning unit, N pixels are sequentially selected from the M pixels in the thinning unit having a pixel value close to the luminance level of the character portion. Pixels are left and other (M−N) pixels are thinned out.

Further, the image obtained by the multi-value reduction processing means is subjected to differential filter processing, and the image output from the differential filter processing section is multiplied by a constant to reduce the multi-value. An image enhancement processing unit that adds to the image obtained by the processing and enhances the image is provided.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described in detail below with reference to the drawings.

Embodiment 1. FIG. 1 shows a configuration of a first embodiment of a binary image multi-value reduction processing apparatus according to the present invention. In this multi-value reduction processing apparatus, the binary image signal input unit 11 is, for example, an image reading device such as a character generation circuit, an image reader, or a facsimile device, and sends the generated or read image as a monochrome binary image. . The data format converter 12 converts the input binary image data format into a multi-valued image data format. For example, as shown in FIG. 2A, the filter determination unit 14 is a two-dimensional filter selection table in which tap lengths and filter coefficient sets set for each tap of the digital filter are stored for each thinning rate. This makes it suitable for storing the signal band corresponding to the thinning rate (reduction rate) input by the thinning rate input unit 13.
The coefficients that realize the dimensional low-pass filter are determined. Note that, for example, as the decimation rate is higher, that is, the reduction rate is higher, a low-pass filter that saves only low-frequency components is selected.

As shown in FIG. 2B, for example, the multi-value processing unit 15 is a low-pass filter such as a Gaussian filter or a wavelet filter as a general digital filter in which a delay circuit, a multiplier and an adder are combined. Is formed by a two-dimensional low-pass filter that is a combination of two in the two-dimensional direction. As a result, the data format conversion unit 12
A two-dimensional low-pass filter process is applied to the binary image signal whose data format has been converted in accordance with the coefficient determined by the filter determination unit 14 to generate a halftone,
A binary image is multivalued to generate a multivalued image. The thinning-out processing unit 16 stores only the pixels located at a constant interval determined by the thinning-out ratio input from the thinning-out ratio input unit 13 among the pixels of the multi-valued image generated by the multi-valued processing unit 15. ,
As a result, a multi-value reduced image is generated. Image display unit 17
Displays the multi-value reduced image generated by the thinning-out processing unit 16 according to the number of gradations on the display.

With such a configuration, the operation actually takes place as shown in FIG. That is, in the binary image signal input unit 11,
2 in black and white with the value "1" or "0" for the scanned image
It is sent as a value image (FIG. 3 (A)). Next, this binary image is converted by the data format conversion unit 12 into the data format of the multivalued image of the value “255” or the value “0” (FIG. 3).
(B)). Subsequently, the multi-valued image is filtered by the multi-valued processing unit 15, and for example, the value “255”, the value “191”, the value “1”.
It is converted into data in five stages of "27", value "63", and value "0" (Fig. 3 (C)).

In FIG. 3C, for simplification, a two-dimensional average value filter for obtaining an average value of a total of 4 pixels of the target pixel and the right side pixel, the lower right side pixel and the lower side pixel of the target pixel to be filtered is shown. Although an example of the operation is shown, since it actually operates as a two-dimensional low-pass filter, the value is different from that in the figure. In this way, a multi-valued low-frequency image composed of data in 5 stages has one pixel in the vertical and horizontal directions when the thinning rate is 4: 1 (reduction rate 1/4) as shown in FIG. 3D, for example. Each image is thinned out and reduced to 1/4 of a pixel, and a multi-value reduced image (see FIG.
(E)).

With such a configuration, the input binary image is converted into the data format of the multivalued image, and the multivalued image is filtered by the two-dimensional low-pass filter determined according to the desired thinning rate. Then, a multi-valued low-frequency image is generated by giving an intermediate gradation, and a multi-value reduced image is generated by thinning out according to a desired thinning rate, and compared with the case where an average value filter or an error diffusion filter is used. Then, since a filter having excellent characteristics as a low-pass filter is used, it becomes possible to perform multi-valued processing that more faithfully reflects the local property of the binary image,
Since it is possible to express with a larger number of halftone levels, the visibility can be significantly improved when a multi-value reduced image is obtained by thinning after binarizing a binary image.

Here, the case where the visibility is improved has been described as the effect of displaying the finally obtained multi-value reduced image on the display, but the present invention is not limited to this.
If the value image is multivalued in advance and stored in the storage device, the display speed on the display can be significantly improved. In fact, if the image for display is hierarchically encoded using Laplacian pyramid encoding, etc., not only the amount of accumulated data can be compressed but also progressive display of the image can be used, so the display speed can be apparently increased. .

Example 2. FIG. 4, in which parts corresponding to those in FIG. 1 are assigned the same reference numerals, shows the configuration of a second embodiment of the binary image multi-value reduction processing apparatus according to the present invention. In the second embodiment, the multi-value quantization processing unit 20 includes a vertical direction one-dimensional filter processing unit 20A and a horizontal direction one-dimensional filter 20B.

In the first embodiment described above, the filter determination unit 1
4 determines the coefficient of the two-dimensional low-pass filter, and the multi-value quantization processing unit 15 realizes multi-value quantization by using the two-dimensional low-pass filter. In the second embodiment, the filter determination unit 21 is used.
Determines the coefficient of the one-dimensional low-pass filter used for multi-value quantization according to the thinning rate (reduction rate). As a result, the multi-value quantization processing unit 20 determines the vertical direction and the horizontal direction determined by the filter determination unit 21 for each of the vertical direction and the horizontal direction with respect to the binary image subjected to the data format conversion by the data format conversion unit 12. Filtering is performed once using the horizontal one-dimensional low-pass filters 20A and 20B.

According to this structure, in addition to the same effect as that of the first embodiment, the vertical and horizontal one-dimensional low-pass filters 20A and 20B are used to perform the filtering process once. The processing can be sped up as compared with the case where the filter processing is performed using the two-dimensional low-pass filter.

Example 3. FIG. 5, in which parts corresponding to those in FIG. 4 are assigned the same reference numerals, shows the configuration of a third embodiment of the binary image multi-value reduction processing apparatus according to the present invention. In the third embodiment, the multi-value quantization processor 30 includes a vertical multi-value quantization processor 30A including a vertical one-dimensional filter processor 30A1 and a vertical one-dimensional thinning processor 30A2, and a horizontal one-dimensional filter processor. 30B1 and a horizontal direction multi-value quantization process 30B including a horizontal direction one-dimensional thinning processing unit 30B2.

In the second embodiment described above, the multilevel halftoning processing section 20.
After the multi-valued low-frequency image is generated in step S1, the thinning-out processing unit 16 performs the thinning-out process to generate the multi-value reduced image. In the second embodiment, first, the input 2 In the vertical direction of the value image, the vertical one-dimensional low-pass filter processing unit 30A1 performs one-dimensional low-pass filter processing, and the vertical multi-valued low-frequency image obtained as a result is subsequently subjected to the vertical one-dimensional thinning processing unit 30A2. Thinning processing is performed only in the direction to generate a vertical multi-value reduced image.

Next, as the horizontal direction multi-value processing 30B, the horizontal one-dimensional low-pass filter processing unit 30B1 performs one-dimensional low-pass filter processing on the horizontal direction of the input vertical multi-value reduced image. For images that are multivalued reduced images in the vertical direction and multivalued low-pass images in the horizontal direction,
Subsequently, the horizontal one-dimensional thinning processing unit 30B2 performs thinning processing only in the horizontal direction, and in this way, multi-value reduction processing is performed in the vertical direction and the horizontal direction to generate a multi-value reduced image having a desired thinning rate. To do.

According to this structure, in addition to the same effect as that of the second embodiment, the vertical multi-value reduction processing is performed, and then the horizontal multi-value reduction processing is performed, so that particularly in the horizontal direction. The number of pixels to be processed can be reduced, and the processing is further speeded up as compared with the case where the vertical and horizontal one-dimensional low-pass filters 20A and 20B are used to perform filter processing once and then thinning processing is performed. can do. In the third embodiment, the filtering process and the thinning process are performed in the order of the vertical direction and the horizontal direction, but the filtering process and the thinning process may be performed in the order of the horizontal direction and the vertical direction. In this way, the processing speed is further increased as compared with the case of the second embodiment.

Example 4. FIG. 6 in which parts corresponding to those in FIG. 1, FIG. 4 and FIG. 5 are assigned the same reference numerals shows a configuration of a fourth embodiment of a binary image multi-value reduction processing apparatus according to the present invention. In the first to third embodiments described above, the filter determination units 14 and 21 determine filters according to the thinning rate input from the thinning rate input unit 13, and the multi-value quantization processing units 15, 20, and 30 perform multi-value quantization. When the processing is performed using the two-dimensional filter, the filtering processing is performed once on the binary image, and when the processing is performed using the one-dimensional filter, the filtering processing is performed once in the vertical direction and once in the horizontal direction, respectively, that is, two times. In the subsequent thinning-out processing unit, if a two-dimensional filter is used in the previous multi-value quantization processing unit, thinning-out processing is performed after filtering processing, and if a one-dimensional filter is used, thinning-out processing is performed after filtering processing in both directions, or The thinning processing is performed once in the vertical and horizontal directions each time the filtering processing in each direction is performed.

On the other hand, in the binary image multi-value reduction processing apparatus of the fourth embodiment, whether the thinning rate input from the thinning rate input unit 13 is the ratio a ^k : 1 (a and k are positive integers). If the thinning-out rate determining unit 41 determines that the thinning-out rate a ^k is not 1, the multi-valued reduction processing selection unit 40 performs the same multi-valued reduction processing as that described in the first to third embodiments. A method is selected, multi-value reduction processing is performed, and a thinning rate a ^k : 1
In this case, a multi-value quantization reduction processing method different from the above-described first to third embodiments is selected.

That is, first, the filter determination unit 42 determines a two-dimensional or one-dimensional filter suitable for thinning out an image to a ratio a: 1, and the two-dimensional or one-dimensional filter determined by the multi-value reduction processing unit 43. Is used to perform a filtering process on a multi-valued image, and then to thin a ratio a: 1 to generate a multi-value reduced image, and this process is recursively repeated k times to finally obtain A multi-value reduced image thinned to a thinning rate a ^k : 1 is generated. The multi-value reduction processing by repeating the filter processing and the thinning-out processing in this way is called a wavelet-type multi-value reduction processing.

Here, when a two-dimensional filter is used as the filter, the two-dimensional low-pass filter processing is performed as in the first embodiment described above, and then the ratio a: 1 is thinned out. When the one-dimensional filter is used as in the second embodiment, first, the one-dimensional low-pass filter is applied in the vertical and horizontal directions of the image, and then the ratio is reduced to a: 1 for reduction. Further, as in the third embodiment, the multi-value reduction processing may be realized by performing one-dimensional filtering processing and thinning processing in the vertical direction and subsequently performing one-dimensional filtering processing and thinning processing in the horizontal direction.

According to such a configuration, in addition to the same effects as those of the above-described first to third embodiments, the wavelet type multi-value reduction processing is performed by repeating the filter processing and the thinning processing. As a result, the processing can be speeded up when the thinning ratio M: N is particularly large at the ratio a ^k : 1.

Example 5. FIG. 7 in which parts corresponding to those in FIG. 6 are assigned the same reference numerals shows the configuration of a binary image multi-value reduction processing apparatus according to a fifth embodiment of the present invention. In the above-described fourth embodiment, the thinning-out rate determining unit 41 determines whether the thinning-out rate is the ratio ak: 1, and the multi-valued reduction processing selecting unit 43 switches the multi-valued reduction processing method according to the result. . On the other hand, in the fifth embodiment, first, the thinning rate M: N input from the thinning rate input unit 13 and the arbitrary thinning rate M: N in the thinning rate division control unit 51 are set to b ^{k ′} : 1 (b ^{k '} <A ^k ) (M,
N, a, k, b, k'are positive integers) and ratio M: Nb ^{k '}
Split into.

Subsequently, the thinning reduction processing of the ratio b ^k : 1 is performed by the wavelet type multi-value reduction processing unit 50 similar to that described in the fourth embodiment.
When b ^{k ′} ≠ 1: 1, further reduction processing (thinning-out processing) is performed on the multi-value reduced image obtained by the wavelet-type multi-value reduction processing unit 50 as described in the first to third embodiments. In the same manner, the multi-value reduction processing unit 53 performs the multi-value reduction processing image with a desired thinning rate (reduction rate) M: N.

According to such a configuration, in addition to the same effects as those of the above-described first to third embodiments, the wavelet type multi-value reduction processing by repeating the filter processing and the thinning processing is performed. Thereby, the processing can be speeded up even when the thinning ratio M: N is not the ratio a ^k : 1.

Example 6. FIG. 8 in which parts corresponding to those in FIG. 7 are assigned the same reference numerals shows the configuration of a sixth embodiment of the binary image multi-value reduction processing apparatus according to the present invention. In the above-described fifth embodiment, the multi-value reduced image generated by the wavelet-type multi-value reduction processing unit 50 is further reduced to realize a desired reduction ratio, and the multi-value reduction shown in the first to third embodiments is performed. Although it is realized by the scaling reduction unit 53, in the sixth embodiment, the nearest neighbor method or the projection method (“High-speed pixel density conversion method based on the projection method”, The Institute of Image Electronics Engineers of Japan, Vol. 11, No. 2 (1982), 72. ~
The additional reduction processing unit 60 is used to generate a multivalued image with a desired reduction ratio through the thinning processing unit 60B.

According to such a configuration, as in the above-described fifth embodiment, the effect of the wavelet-type multi-value reduction processing is used, and the nearest neighbor method or the projection method is used as a part of the filter processing. The processing can be simplified as compared with the fifth embodiment. In Examples 1 to 6 described above, a two-dimensional filter having a high degree of freedom in filter design is used when image quality is given the highest priority (Example 1), and a one-dimensional filter is used when processing speed is given priority. (Examples 2 and 3), and further, wavelet type filter processing using a one-dimensional filter is used (Examples 4 to 6), depending on the operating environment and the purpose of use of the multi-value reduction processing. The processing method can be selected.

Example 7. FIG. 8 in which parts corresponding to those in FIG. 1 are assigned the same reference numerals shows a configuration of a binary image multi-value reduction processing apparatus according to a seventh embodiment of the present invention. In the above-described first to sixth embodiments, in the multi-valued low-frequency image generated by the multi-valued processing unit, pixels that are periodically positioned at regular intervals that match the thinning-out ratio are stored, and there is a conventional thinning-out process. However, such a method could not solve the problem that information in the thin line portion below the thinning rate is missing.

On the other hand, in the seventh embodiment, the fine line determination unit 7
The thickness of the line at each pixel position of the multi-valued image input at 1 is measured, and if the measured value is less than or equal to the thinning rate, this is determined as a thin line, and the determination result by the thin line determination unit 71 is a thin line. Only when thinning out positions including pixels as thinning units, the thinning-out method is adaptively changed for each position to realize multi-value reduction processing of a binary image in which information loss in the thin line portion is suppressed. .

In practice, the thin line determination section 71 determines whether, for each pixel of the multi-valued image, it belongs to a thin line whose vertical line width is less than or equal to the thinning rate, or whose horizontal line width is less than or equal to the thinning rate. If the line width in at least one of the directions is equal to or smaller than the thinning rate, it is determined that the pixel belongs to a thin line. Note that the order of checking the line width of each pixel in each direction may be the order of the horizontal direction and the vertical direction instead of the order of the vertical direction and the horizontal direction shown here.

In addition, the adaptive thinning-out processing unit 70B, in the case where the pixels determined to belong to the thin line by the thin line determining unit 71 when performing the thinning-out are included in the minimum unit of thinning-out, for each minimum unit of thinning-out, Adaptive thinning processing is performed such that M pixels are left in order from the pixel value closest to the pixel value representing the line segment. For example, when the thinning rate is M: N, the adjacent N pixels are the minimum unit of thinning, and when the pixel value representing the line segment is “0” and the background pixel value is “255”, the pixel value is the minimum. Leave M pieces in order from the one.

Here, as shown in the first embodiment, the case where the binary image is subjected to the multi-valued processing by the two-dimensional filter, and as shown in the second embodiment, the one-dimensional filter is moved in the vertical and horizontal directions before the thinning processing. In the case of carrying out multi-value quantization by applying it, the minimum unit of thinning (thinning ratio M: N) is a two-dimensional square block N × N.
It is a pixel and performs adaptive thinning processing two-dimensionally. On the other hand, when thinning is performed every time the one-dimensional filter is applied in the vertical or horizontal direction as shown in the third embodiment, the minimum unit of thinning (thinning ratio M: N) is one-dimensional adjacent N pixels and one-dimensional Adaptively thinning out.

Actually, FIG. 10A shows a method of thinning out pixels which are periodically positioned at intervals corresponding to the thinning rate, and FIG. 10B shows an adaptive thinning method for each thinning unit. I will show you how to change to. In the figure, 90A, 90
B shows an example of a unit of thinning, 90C shows a pixel saved by the periodic thinning process, and 90D shows a pixel saved by the adaptive thinning process.

FIG. 11 shows a thin line determination processing procedure executed by the thin line determination section 71, and FIG. 12 shows an adaptive thinning processing procedure performed by the adaptive thinning processing section. That is, in the thin line determination processing procedure (FIG. 11), from step ST0, in step ST1, it is determined whether or not the pixel belongs to a line segment, and if a positive result is obtained, the process moves to step ST2 and the vertical direction in which the pixel is included. Line width is measured and the next step ST
In 3, it is judged whether the line width is less than the thinning rate. If an affirmative result is obtained here, the process proceeds to step ST4, it is determined that the pixel belongs to the thin line, and the process proceeds to step ST7.

On the other hand, if a negative result is obtained in step ST3, the horizontal line width at the position including the pixel is measured in step ST5, and it is determined in the next step ST6 whether the line width is less than the thinning rate. To do. If a positive result is obtained here, the process proceeds to step ST4, it is determined that the pixel belongs to the thin line, and the process proceeds to step ST7. In addition, when the negative result is obtained in step ST1 and the negative result is obtained in step ST6, the process directly proceeds to step ST7. Step S
At T7, it is determined whether or not all pixels have been determined, and if a negative result is obtained, the process returns to step ST1 described above and the above processes are repeated, and if a positive result is obtained, the process proceeds to step ST8, and the fine line determination processing procedure. To finish.

Further, in the adaptive thinning-out procedure (FIG. 12), it is entered from step ST10, it is judged in step ST11 whether or not there is a pixel determined to be a thin line in the thinning-out unit, and if a positive result is obtained, step ST12. In, the pixel having the smallest pixel value in the thinning unit is left, the other pixels are thinned, and the process proceeds to step ST14. When a negative result is obtained in step ST11 described above, the process proceeds to step ST13 to perform normal thinning processing, and then proceeds to step ST14.
In step ST14, it is determined whether or not all the thinning-out units have ended. If a negative result is obtained, the process returns to step ST11 to repeat the above-described processing, and if a positive result is obtained, step ST15.
Then, the adaptive thinning-out processing procedure ends.

According to such a configuration, the same effects as those of the above-described first to sixth embodiments can be obtained, and when the thinning process is performed on the image, the pixel position determined to be a thin line in the binary image is obtained. Since adaptive thinning-out processing is performed in order to suppress the information loss in the above, the loss of the thin line portion can be effectively prevented and the image quality can be significantly improved.

Example 8. FIG. 13 in which parts corresponding to those in FIG. 1 are assigned the same reference numerals shows the configuration of an embodiment 8 of the binary image multi-value reduction processing apparatus according to the present invention. Example 1 described above
7 to 7 have described the case where the multi-value reduction processing of the binary image is performed by the filter processing and the thinning processing (adaptive thinning processing), the multi-value processing (low-pass filter processing) which is a kind of blurring operation is described. In the process, blurring may occur even in a portion where the occurrence of blurring is not necessarily desired in the reduced image, and this does not solve the problem that the image feels totally blurred.

On the other hand, in the eighth embodiment, the differential filter processing unit 83 performs a differential filter on the multi-value reduction processing result (image) 81 which has been subjected to the multi-value reduction processing according to any of the first to seventh embodiments. Then, the emphasis parameter setting unit 84 determines the degree to which the image is emphasized, and the multiplication circuit 85 multiplies the differential image obtained by the differential filter processing 83 by the value determined by the emphasis parameter setting unit 84 to add the addition circuit. At 86, the multiplication result of the multiplier 85 and the multi-value reduction processing result (image) 81 are added to eliminate excessive blurring remaining in the multi-value reduced image.

According to such a configuration, the filtering process,
Since the multi-value reduction processing image obtained by the adaptive thinning processing is subjected to the emphasis processing by the differential filter processing to reduce the blur portion, there is an effect that the multi-value reduction processing with higher visibility can be realized.

[0053]

As described above, according to the present invention, an input binary image is converted into a multivalued image data format, and this multivalued image is reduced at a reduction ratio N / M (N and M are positive integers). ), A low-pass filter determined according to the thinning rate M: N is applied to generate a multivalued low-frequency image by giving an intermediate gradation, and thinning is performed according to a desired thinning rate M: N to obtain a multivalued image. By generating a reduced image, it is possible to realize a binary image multi-value reduction processing apparatus that can significantly improve the visibility when a multi-value reduced image is obtained by thinning after binarizing a binary image. .

The thinning ratio M: N is the ratio a ^k : 1.
In the case of (N, M, a, and k are positive integers), the multi-valued image is filtered by the low-pass filter determined according to the thinning rate a: 1, and the thinning rate a: 1 is set for each filtering process. the process of thinning recursively repeated k times, finally thinning index a ^k: by outputting a multi-valued reduced image, the thinning ratio is the ratio a ^k: 1, then a simple structure and fast It is possible to realize a binary image multi-value reduction processing apparatus that can significantly improve the visibility when processing a binary image to obtain a multi-value reduced image.

If the thinning ratio M: N is not the ratio a ^k : 1, the multi-valued image is first thinned out by the ratio b ^{k ′} : 1 (b ^{k ′} <a.
^k ) (M, N, a, k, b, k ′ are positive integers) and the process of thinning is recursively repeated k ′ times, and for the multi-value reduced image obtained as the thinning result , As a thinning process, the thinning rate M: Nb ^{k '}
The thinning-out processing is performed in step S3, and finally the multi-value reduced image thinned out to the thinning-out ratio M: N is output, so that even if the thinning-out ratio is not the ratio a ^k : It is possible to realize a binary image multi-value reduction processing apparatus that can significantly improve the visibility when multi-valued a value image to obtain a multi-value reduced image.

Further, the thinning rate M: N is the ratio a ^k :
When it is not 1, first, the multi-valued image is thinned out at a bk ^' : 1 (b
^{k '} <a ^k ) (M, N, a, k, b, k'are positive integers)
The filtering process and the thinning process are recursively repeated k ′ times, and the thinning-out result obtained by the thinning process is continuously thinned by the nearest neighbor method or the projection method at a thinning rate M: Nb ^{k ′.} By outputting the multi-value reduced image which has been thinned out and finally thinned to the thinning ratio M: N, even if the thinning ratio is not the ratio a ^k : 1, it is processed with a simple structure and at high speed to obtain a large number of binary images. The visibility when binarizing to obtain a multi-value reduced image can be significantly improved.

Further, in the case where a portion corresponding to a thin line having a thinning rate or less is determined in the input binary image and the result obtained by performing the low-pass filter processing on the portion determined to be the thin line is thinned out, the thinning method is locally performed. By adaptively changing the information, even when the information of the thin line portion below the thinning rate exists in the binary image, it is possible to effectively prevent the information of the thin line portion from being lost. It is possible to realize a binary image multi-value reduction processing apparatus that can remarkably improve the visibility when converted into a multi-value reduced image.

Further, when further determining the thin line, it is determined whether or not each pixel of the input binary image belongs to a line segment having a width equal to or less than the thinning rate in each of the vertical and horizontal directions.
If the line width in either direction is less than or equal to the thinning rate, it is possible to reliably determine the thin line portion in the binary image that is less than or equal to the thinning rate by determining the corresponding pixel as the line segment portion. It is possible to realize a binary image multi-value reduction processing apparatus capable of significantly improving the visibility when binarizing to obtain a multi-value reduced image.

Further, in the case of thinning out to a thinning rate M: N, with respect to pixels which are not determined to be a thin line portion, only pixels which are periodically positioned at intervals matching the thinning rate M: N are left and it is determined to be a thin line portion. In the place where the thinning-out unit is included in the thinning-out unit, N pixels are left in order from the M pixels in the thinning-out unit having a pixel value close to the luminance level of the character portion, and the other (M−N) pixels are thinned out. As a result, it is possible to realize a binary image multi-value reduction processing apparatus capable of more accurately reflecting thin line information and adaptively thinning.

Further, the multi-value reduced image obtained by the multi-value quantization processing means and the thinning-out processing means is multiplied by the constant image of the output obtained by the differential filter processing and added to the multi-value reduced thumbnail image. However, by emphasizing the same image, the excessive blur part mixed in the image in the process of the multi-value reduction processing is emphasized and reduced, and the visibility when the multi-valued binary image is obtained to obtain the multi-value reduced image. It is possible to realize a binary image multi-value reduction processing apparatus capable of significantly improving the property.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a first embodiment of a binary image multi-value reduction processing apparatus according to the present invention.

FIG. 2 is a schematic diagram used for explaining a configuration of a filter determination unit and a multi-value conversion processing unit of the binary image multi-value reduction processing apparatus of FIG.

FIG. 3 is a schematic diagram for explaining an operation of the binary image multi-value reduction processing apparatus of FIG.

FIG. 4 is a block diagram showing a configuration of a second embodiment of a binary image multi-value reduction processing apparatus according to the present invention.

FIG. 5 is a block diagram showing a configuration of a binary image multi-value reduction processing apparatus according to a third embodiment of the present invention.

FIG. 6 is a block diagram showing a configuration of a fourth embodiment of a binary image multi-value reduction processing apparatus according to the present invention.

FIG. 7 is a block diagram showing a configuration of a binary image multi-value reduction processing apparatus according to a fifth embodiment of the present invention.

FIG. 8 is a block diagram showing the configuration of a sixth embodiment of a binary image multi-value reduction processing apparatus according to the present invention.

FIG. 9 is a block diagram showing a configuration of a seventh embodiment of a binary image multi-value reduction processing apparatus according to the present invention.

10 is a schematic diagram for explaining the operation of the adaptive thinning-out process in the binary image multi-value reduction device of FIG.

FIG. 11 is a flowchart showing an operation of a thin line determination unit of the binary image multi-value reduction device of FIG.

FIG. 12 is a flowchart showing the operation of the adaptive thinning processing unit of the binary image multi-value reduction device of FIG.

FIG. 13 is a block diagram showing Embodiment 8 of the binary image multi-value reduction processing apparatus according to the present invention.

FIG. 14 is a block diagram showing a configuration of a reduced character display device as a conventional binary image multi-value reduction processing device.

[Explanation of symbols]

1 Character generation circuit 2 Filter circuit 3 Display device 11 Binary image signal input unit 12 Data format conversion unit 13 Decimation rate input unit 14, 21, 42, 52A, 52B, 72 Filter determination unit 15, 20, 70A Multi-valued processing Parts 16, 26, 60B Thinning processing unit 17 Image display units 20A, 30A1 Vertical direction one-dimensional filter processing units 20B, 30B1 Horizontal direction one-dimensional filter processing units 30, 43, 53, 70 Multi-value reduction processing unit 30A Vertical direction multi Quantization processing unit 30B Horizontal-direction multi-value processing unit 30A2 Vertical-direction one-dimensional thinning processing unit 30B2 Horizontal-direction one-dimensional thinning processing unit 40 Multi-value reduction processing selection unit 41 Thinning rate determination unit 50 Wavelet-type multi-value reduction processing unit 51 thinning rate division control unit 60 additional reduction processing unit 60A nearest neighbor method or projection method 70B adaptive thinning processing unit 71 thin line determination unit 81 multi-value reduction Processing result 82 Image enhancement processing unit 83 Differential filter processing 84 Enhancement parameter setting unit 85 Multiplier 86 Adder

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI G09G 5/36 520J 520P H04N 1/40 103B (56) Reference JP-A-5-328106 (JP, A) JP-A-5- 40826 (JP, A) JP-A-5-89238 (JP, A) JP-A-7-95394 (JP, A) JP-A-2-268073 (JP, A) JP-A-3-34677 (JP, A) JP-A-8-191387 (JP, A) JP-A-3-259193 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G09G 5/00 G09G 5/36 G09G 5/391 H04N 1/393 H04N 1/40

Claims (10)

(57) [Claims] 1. A data format conversion means for converting an input binary image into a data format of a multivalued image, and a reduction rate N / M.
(N, M is a positive integer) the desired decimation factor comprised in M: according to N, the low-pass filter for band-limiting the multi-value image
Filter determining means for determining the tap length and the filter coefficient, and the multi-valued image obtained by the data format converting means is filtered by the low-pass filter determined by the filter determining means to extract a low-pass signal component, The multi-valued low-frequency image obtained by the multi-valued low-frequency image and the multi-valued low-frequency image by giving an intermediate gradation to the multi-valued image and the multi-valued low-frequency image is set to the desired thinning rate M: N. Thinning processing means for thinning according to the generation of a multi-value reduced image,
The multi-value processing means is obtained from the data format conversion means.
To the multi-valued image that is
Vertical one-dimensional low-pass filter that filters horizontally
Filter processing unit and horizontal one-dimensional low-pass filter processing unit
At the same time, the filter determination means causes the thinning rate M:
As N becomes larger, the above vertical one-dimensional low-pass filter
Processing unit and the horizontal one-dimensional low-pass filter processing unit
To increase the tap length of the one-dimensional low-pass filter of
A binary image multi-value reduction processing apparatus characterized by being present. 2. The input binary image is converted into multi-valued image data.
Data format conversion means for converting to format and reduction rate N / M
(N and M are positive integers) according to the desired thinning rate M: N.
The low-pass filter that limits the band of the multi-valued image.
Filter determining means for determining tap length and filter coefficient
And the multivalued image obtained by the data format conversion means
Is the low-pass filter determined by the filter determining means.
Filter to extract the low-frequency signal component,
When a multi-valued low-frequency image is generated by giving an intermediate gradation to a multi-valued image
In both cases, the multi-valued low-frequency image is subjected to the desired thinning rate M: N.
Multi-value reduction processing that thins out according to
The multi-value reduction processing means is provided with
The multi-valued image obtained by the
Vertical one-dimensional low-pass fill that filters in the direction
Data processing unit and the above vertical direction one-dimensional low-pass filter processing unit
Vertical processing to perform vertical thinning processing on these image data
Vertical multi-valued processing unit including unidirectional one-dimensional thinning processing unit
And a horizontal one-dimensional low-pass that performs filtering in the horizontal direction
Filter processing unit and the horizontal one-dimensional roper Sfilter processing
Performs horizontal decimation processing on image data from the science department.
A horizontal direction multi-value processing unit including a horizontal direction one-dimensional thinning processing unit
The filter determining means is provided with
As the ratio M: N increases, the above-mentioned one-dimensional vertical low-pass
Filter processing unit and the above horizontal one-dimensional low-pass fill
Increase the tap length of the one-dimensional low-pass filter in the data processing unit
Binary image multi-value reduction processing characterized by
apparatus. 3. The one-dimensional low-pass filter is a Gaussian filter.
It is an unfilter, and the filter determination means is 2:
Corresponding to the decimation rate of 1, 4: 1, 8: 1 respectively,
The tap length of the one-dimensional low-pass filter is set to 3, 5, 9,
The filter coefficients are (1.0, 2.0, 1.0), (1.
0, 1.5, 2.0, 1.5, 1.0), (1.0,
1.5, 2.0, 2.5, 3.0, 2.5, 2.0,
1.5, 1.0).
Is a binary image multi-value reduction processing apparatus according to claim 2. 4. The input binary image is converted into multi-valued image data.
Data format conversion means for converting to format and reduction rate N / M
(N and M are positive integers) desired thinning rate M: N
Rate a ^k  Thinning out to determine whether or not: 1 (a and k are positive integers)
And the determination result of the thinning rate determining means.
The above thinning ratio M: N is the ratio a ^k  : 1 in case of above multi-value
A band-limiting low for thinning out images at a thinning rate a: 1
Filter determining means for determining a pass filter, and
The low-pass filter determined by the filter determination means
Multi-valued processing for applying the filter processing to the multi-valued image
Every time the filtering process is performed, the thinning rate a: 1 is used.
Repeat thinning processing and k times each, and finally thin
Rate a ^k Multi-value reduction that outputs a multi-value reduced image thinned to 1
Binary image multi-value conversion characterized by comprising small processing means
Reduction processing device. 5. The thinning rate M: N is defined as the thinning rate.
b ^{k ’} : 1 (b ^{k ’} <A ^k ) And thinning rate M: Nb ^{k ’} Split into
The thinning rate dividing control means
As the determination result of the step, the thinning ratio M: N is the ratio a. ^k  :
If it is not 1, the multi-value reduction processing means first
Multi-valued image is thinned out b ^{k ’} : 1 (b ^{k ’} <A ^k  ) (B,
k'is a positive integer)
For the obtained multi-value reduced image, the thinning rate M:
Nb ^{k ’} The thinning process is performed with Thinning rate M: N
A multi-value reduced image that has been thinned out
The binary image multi-value reduction processing apparatus according to claim 4. 6. The thinning rate M: N is the thinning rate
b ^{k ’} : 1 (b ^{k ’} <A ^k ) And thinning rate M: Nb ^{k ’} Split into
The thinning rate dividing control means
As the determination result of the step, the thinning ratio M: N is the ratio a. ^k  :
If it is not 1, the multi-value reduction processing means first
Multi-valued image is thinned out b ^{k ’} : 1 (b ^{k ’} <A ^k  ) (B,
k'is a positive integer)
For the obtained multi-value reduced image, the nearest neighbor method or
Thinning rate M: Nb by projection method ^{k ’} Thinned out with
Finally, the multi-value reduced image thinned to the thinning ratio M: N
5. The binary image multivalued according to claim 4, which is output.
Quantization / reduction processing device. 7. A thin line having a thinning rate equal to or less than a thinning rate in the binary image.
The thin line determination means for determining the location and the thin line determination means
The low-pass filter is applied to the parts that are determined to be thin lines.
When thinning the result of applying
And adaptive thinning processing means for adaptively changing
5. The method according to any one of claims 1 to 4, characterized in that
Binary image multi-value reduction processing device. 8. The thin line determination means is provided for each of the binary images.
Pixels have a width equal to or less than the thinning rate in each of the vertical and horizontal directions
Determine if it belongs to a line segment,
If the line width is less than the thinning rate, select the corresponding pixel
The binary image according to claim 7, wherein
Multi-value reduction processor. 9. The adaptive thinning processing means is a thinning rate.
Pixels that are not determined as thin line parts when thinning out to M: N
About the thinning rate M: N, periodically.
The thinning process is performed so that only the pixels to be placed remain.
To the part that includes the pixels determined to be minutes in the thinning unit
As for the thin line portion of N pixels in the above thinning unit
M pixels in order from a pixel value close to the brightness level of
The remaining (N−M) pixels are thinned out.
8. The binary image multi-value reduction processing device according to claim 7. 10. Obtained by the multi-value reduction processing means
Differential filtering to apply differential filtering to the captured image
And the image output from the differential filter processing unit
Multiply by a constant and add to the image obtained by multi-value reduction processing
The same image And a processing unit for enhancing an image to be adjusted
5. The method according to any one of claims 1 to 4, characterized in that
Binary image multi-value reduction processing device.