JP3990103B2 - Image processing apparatus and image processing method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing apparatus such as a digital copying machine and a facsimile machine capable of performing image processing for improving the image quality of an output image. In particular, the present invention relates to various types such as character originals, halftone originals, and character originals on halftone dots. The present invention relates to an image processing apparatus capable of performing suitable filter processing on a document.
[0002]
[Prior art]
In a conventional digital copying machine, an edge amount (a data value change amount in a position direction on an image surface) is calculated based on local information (for example, a change characteristic of an image data value) of a document in which characters and patterns are mixed, Appropriate filtering processing is performed on each area on the image plane.
For example, in the conventional technique disclosed in Japanese Patent Publication No. 6-5879, the output of the smoothing filter and the edge enhancement filter is mixed based on the edge detection level information using an edge detection filter. In other words, the flat part where the image data value changes gently is subjected to smoothing filtering that removes minute dirt and smoothes the change so that the image data value of the background such as the character part changes to the image data value of the character part. By performing filtering that emphasizes the edge at the edge portion where the value changes abruptly, good filtering processing is performed in any portion of the image in which characters and patterns are mixed. In this prior art, the edge detection filter is used to suppress the occurrence of moire by using a filter that has a characteristic that it is difficult to detect a halftone dot as an edge.
In addition, the prior art disclosed in Japanese Patent Application Laid-Open No. 7-95409 is provided with an image area separation circuit that separates a character area and a non-character area with high accuracy, and performs optimum processing based on the separation information. Even when a character is not detected as an edge portion, the above-mentioned problem is solved by performing adaptive edge enhancement (adaptive edge filtering processing) on the processing for non-character after smoothing filtering.
[0003]
[Problems to be solved by the invention]
However, in the prior art disclosed in Japanese Patent Publication No. 6-5879, the frequency transfer characteristics of the smoothing filter and the edge emphasis filter are greatly different from each other. There is a problem that unnatural image quality deterioration is caused at the filter switching portion. In other words, since a large amount of edge is not detected for a character on a halftone dot even though it is a character part, smoothing filtering is performed for such a region because it is difficult to separate the character part with high accuracy. It will end up. Edge images such as characters must always be detected as edge portions. In reality, a highly accurate detection mechanism or filter is required, which is disadvantageous in terms of cost.
In the prior art disclosed in Japanese Patent Laid-Open No. 7-95409, the image area separation circuit is large-scale, which is disadvantageous in terms of cost.
The object of the present invention is to solve such a problem of the prior art, and according to the feature amount of an image without using an image region discriminating unit such as an image region separating unit which has a large hardware scale and increases costs. By processing, for images with mixed text, halftone dots, and flat areas, edge images such as characters and lines are high-quality while suppressing the generation of moire due to halftone dots in areas with large edge amounts. An object of the present invention is to provide an image processing apparatus which can be reproduced and can suppress the occurrence of unnatural image quality deterioration at the switching portion.
[0004]
[Means for Solving the Problems]
In order to solve the above-mentioned problem, according to the first aspect of the present invention, in an image processing apparatus capable of performing conversion according to frequency on input image data by a spatial filter, an image from multi-tone input image data is converted into an image. an edge amount detection means for detecting the surface of a plurality of directions of edge amount, respectively, and the edge direction characteristic determining means for comparing each direction edge amount mutually detected by the edge amount detection unit, the edge direction characteristic determination means As a result of the comparison, a converted edge amount is calculated by multiplying a constant equal to or less than 1 when two oblique directions are determined to have strong edge direction characteristics, and multiplying by 1 in other cases. One edge amount conversion means and a non-linear conversion in which the output edge amount is saturated with respect to the post-conversion edge amount that is greater than or equal to a predetermined amount output from the first edge amount conversion means A second edge amount conversion means having a gender, and a first filtering means for applying a conversion corresponding to the frequency relative to the input image data, and said first filtering means to the input image data different, the second filter processing means for performing conversion in accordance with the frequency, a multiplier for multiplying the output result of the second edge amount conversion means according to the output result and the second filtering means, the multiplier And an adder for adding the output result of the first filter processing means, the first filter processing means enhances only the data component of a specific frequency region, and the spatial frequency transfer characteristic is increased. A band-pass filter that corrects the frequency transfer characteristic so as to attenuate the signal having the frequency transfer characteristic in the oblique 45-degree direction in the two-dimensional space shown in FIG.
According to a second aspect of the present invention, there is provided an image processing method capable of performing conversion according to frequency on input image data by a spatial filter. An edge amount detection step for detecting each amount, an edge direction characteristic determination step for comparing edge amounts in each direction detected by the edge amount detection step, and a comparison result of the edge direction characteristic determination step A first edge amount conversion step of calculating a converted edge amount by multiplying a constant of 1 or less when both of the oblique directions are determined to have strong edge direction characteristics, and multiplying by 1 in other cases; A non-linear change in which the output edge amount is saturated with respect to the post-conversion edge amount equal to or greater than the predetermined amount output from the first edge amount conversion step. The second edge amount converting step for performing the above, the first filter processing step for performing the conversion according to the frequency for the input image data, and the first filter processing step for the input image data are different. A second filter processing step that performs conversion according to frequency, a multiplication step that multiplies the output result of the second edge amount conversion step and the output result of the second filter processing step, and An addition step of adding the output result and the output result of the first filter processing step, wherein the first filter processing step enhances only a data component in a specific frequency region, and improves a spatial frequency transfer characteristic. A bandpass that corrects the frequency transfer characteristic so as to attenuate a signal having a frequency transfer characteristic of 45 degrees obliquely in the two-dimensional space shown. Characterized in that it is a step of filtering.
[0006]
[Action]
According to the above-described means, in the first and second aspects of the invention, the amount of edge in a plurality of directions of the image plane from the multi-tone input image data , that is, the vertical, horizontal, left-up diagonal, and right-up diagonal When the edge amounts in the four directions are detected and the detected edge amounts in the respective directions are compared with each other, if it is determined that the two oblique directions have strong edge direction characteristics, 1 or less. Otherwise, the first edge amount conversion by multiplying by 1 is performed, and when the converted edge amount by the first edge amount conversion is input, the converted edge amount equal to or greater than a predetermined amount Is subjected to a second edge amount conversion in which the output edge amount is saturated, the input image data is subjected to a first conversion corresponding to a frequency, and the input image data , To a different frequency than the first conversion Second transform is applied with Flip, the second output result of the output result and the second converted by the edge amount conversion is multiplied by the output result and the output result is the addition of the first conversion of the multiplication is, at that time, in the first conversion, only the data components in a specific frequency region is enhanced, the signal having the frequency transfer characteristics of the oblique 45 degree direction on the two-dimensional space showing the spatial frequency transfer characteristic attenuation To do.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing the configuration of an image processing apparatus such as a digital copying machine in which the present invention is implemented. As shown in the figure, in this image processing apparatus, the spatial frequency characteristics of the image data input by the image input unit 1 are corrected by the filter processing unit 2 and further input to the density conversion unit 3. Then, the density conversion unit 3 performs conversion so that the gradation characteristics of the output image from the image output unit 5 with respect to the document become the desired gradation characteristics. Next, the halftone processing unit 4 performs pseudo halftone processing, and the image output unit 5 outputs an image to paper or the like.
The image processing apparatus according to the present invention relates to the filter processing unit 2 among them.
FIG. 2 is a configuration block diagram of a main part of the image processing apparatus showing an example of the embodiment of the present invention. As illustrated, the edge amount detection unit 11, the edge direction characteristic determination unit 12, the first edge amount conversion unit 13, the second edge amount conversion unit 14, the first filter processing unit 15, and the second filter processing unit. 16 etc. are provided. Note that the edge amount detection means, the edge direction characteristic determination means, the first edge amount conversion means, the second edge amount conversion means, the first filter processing means, and the second filter processing means described in the claims include: In the embodiment, in that order, the edge amount detection unit 11, the edge direction characteristic determination unit 12, the first edge amount conversion unit 13, the second edge amount conversion unit 14, the first filter processing unit 15, and the first filter processing unit 15, respectively. This is realized by two filter processing units 16.
With this configuration, in this image processing apparatus, input image data is input to the first filter processing unit 15, and the first filter processing unit 15 performs a convolution operation with a predetermined filter coefficient, depending on the frequency. Correction is performed. The filter applied here is, for example, a spatial filter having coefficients as shown in FIG.
[0009]
The matrix shown in FIG. 3 corresponds to pixels arranged vertically and horizontally, and the pixel of interest (target for which a data value is to be obtained) corresponding to the center of the matrix (indicated by the value 32 in the example shown in the figure). When determining the data value of the pixel), not only the value of the pixel of interest having the coefficient shown in the figure, but also the values of the pixels around the pixel of interest are added by multiplying the signed coefficient as shown in the figure. To reflect. Note that the coefficient of each pixel shown in FIG. 3 is the frequency transfer characteristic in the main scanning direction (for example, the horizontal direction) and the sub-scanning direction (for example, the vertical direction) when scanning the image plane, and the data component in the 45 degree (diagonal) direction. It is decided in consideration of weakening.
With such a coefficient, as shown in FIG. 4, this filter emphasizes an intermediate frequency band and does not emphasize only the 45 degree direction in the two-dimensional space (on the image plane). Band-enhanced frequency transfer characteristics can be realized. As a result, a halftone image with high density line number and 45 degree screen angle (an angle of 45 degrees with respect to the vertical or horizontal direction of the image surface such as a screen or paper) (for example, in the case of a monochrome original, generally This makes it possible to perform moderate edge emphasis processing in other areas while suppressing moiré at halftone dots arranged in a 45 degree direction.
Also, the input image data is input to the edge amount detection unit 11, and the edge amount included in the image data is detected. The edge amount is a data change amount (data differential value) in the position direction of the image plane. For example, this value becomes large at the boundary between the background portion and the character portion. FIG. 5 shows a configuration example of the edge amount detection unit 11.
[0010]
The four-direction edge detecting means shown in FIG. 5 is constituted by a first-order differential filter as shown in FIG. Among these, FIG. 6A shows the filter coefficient of the A block shown in FIG. 5, detects the edge amount (edge component) in the vertical direction in the image, and outputs the absolute value a of the edge amount. FIG. 6B shows the filter coefficient of the B block shown in FIG. 5, detects the edge amount in the horizontal direction, and outputs the absolute value b of the edge amount. 6 (c) and 6 (d) are filter coefficients of the C and D blocks shown in FIG. 5, respectively, and detect edge amounts in the diagonally upward direction and diagonally upward direction. d is output respectively.
The edge direction characteristic determination unit 12 receives the edge amount, compares the edge amounts a, b, c, and d, ranks them, and determines the directionality or characteristics of the edge image based on the ranks.
An example of a processing flow in the edge amount detection unit 11, the edge direction characteristic determination unit 12, and the first edge amount conversion unit 13 is shown in FIG.
First, as described above, the edge amount detection unit 11 detects the edge amounts a, b, c, and d (S1). Then, the edge direction characteristic discriminating unit 12 receives the edge amount, compares the edge amounts a, b, c, and d, ranks the edge amounts in the four directions (S2), and rises left and right. It is determined whether or not the edge amounts c and d in the diagonal direction are the maximum and second largest values (S3). A 45-degree screen angle halftone image in which both left-up diagonal and right-up diagonal directions show strong edge components is highly likely to satisfy this determination condition.
Subsequently, the first edge amount conversion unit 13 determines the edge amount with respect to the maximum value of the edge amounts a, b, c, and d in each direction based on the determination result in the edge direction characteristic determination unit 12. Increase / decrease and output the conversion edge amount E. In the processing flow shown in FIG. 7, when both the left-up diagonal and right-up diagonal directions indicate strong edge components (Yes in S3), a constant α (α: a constant from 0 to 1, for example, 0.8). Is multiplied to reduce the edge amount and output (S4). It is highly possible that the image has a strong edge component in both the left-upward diagonal direction and the right-upward diagonal direction, so there is a high possibility that it is a halftone dot image. It suppresses the occurrence of moiré.
As described above, when the processing flow shown in FIG. 7 is executed, an edge amount having a value lower than the original edge amount is output for a 45 degree screen angle halftone image that causes moire. Generation of moire due to emphasis can be reduced, and edge emphasis can be performed on characters and the like.
[0011]
Next, the second edge amount conversion unit 14 that has received the edge amount E converts the edge amount E into an edge amount F. The edge amount conversion unit 14 is a conversion table that realizes conversion as shown in FIG. 8, for example, and has a nonlinear conversion characteristic such that the converted edge amount F is saturated with respect to an edge amount E that is greater than or equal to a predetermined amount. Have.
On the other hand, in the second filter processing unit 16, a predetermined frequency component of the input image data is extracted by, for example, a secondary differential filter as shown in FIG. The second-order differential filter shown in FIG. 9 can realize the frequency transfer characteristic as shown in FIG. 10 and extracts a high frequency component. Further, the multiplier 17 multiplies the frequency component extracted by the second filter processing unit 16 and the edge amount F from the second edge amount conversion unit 14 and outputs the result to the adder 18. That is, the output of the second edge amount conversion unit 14 is valid only for data components having a high frequency band. The adder 18 adds the output value of the multiplier 17 and the output value of the first filter processing unit 15 and outputs image data.
In this embodiment, an example of a 5 × 5 size primary differential filter is shown as the edge amount detection unit 11, but the present invention is not limited to this, and a different size may be used, or a secondary differential filter. But you can. In either case, almost the same effect can be obtained.
In this embodiment, the two first-order differential filters C and D in the diagonal direction of the edge amount detection unit 11 are shown as having an angle of 45 degrees to the left and 45 degrees to the right. However, the present invention is not limited to this. The filter may detect a characteristic at an angle different from that. It is changed according to the angle of edge information to be detected.
Further, in this embodiment, the second edge amount conversion unit 14 is shown as an example in which the edge amount E is nonlinearly converted by the conversion table, but may be realized by a calculation formula or the like.
[0012]
With the configuration described above, the first filter processing unit 15 having the characteristics of the band enhancement filter is used for a region where the edge amount detected in the realization method is small, such as a halftone part having a high line number density. Therefore, it is possible to perform correction with improved sharpness of the pattern portion while suppressing moire. For edge portions other than halftone dots, an adaptive edge enhancement process based on the edge amount detected and calculated in the data flow from the edge amount detection unit 11 to the second edge amount conversion unit 14 is performed. Image reproduction with improved sharpness of the image portion.
A configuration without the second edge amount converter 14 is also possible. However, in the configuration including the second edge amount conversion unit 14, the output edge amount can be suppressed within a predetermined amount, and thus it is possible to prevent the output result from being shaken by an abnormally large edge amount.
In the mixing of both the smoothing filter and the emphasis filter output as described in Japanese Patent Publication No. 6-5879, the frequency transfer characteristics of both filters are greatly different. The switching portion is unnatural, causing image quality degradation. However, in the embodiment of the present invention, as described above, both the band enhancement filter and adaptive edge enhancement filter outputs are configured to be mixed. In the middle frequency band, all outputs have emphasis characteristics, and there is no image degradation at the switching portion.
In the prior art described in Japanese Patent Laid-Open No. 7-95409, adaptive edge emphasis is performed after the smoothing filter, and there is no image quality deterioration at the boundary portion, but it is costly to satisfy the emphasis on the character portion. Therefore, it is necessary to have an image area separation circuit that separates a character area and a non-character area which are disadvantageous to the high accuracy. In the present invention, sufficient edge emphasis can be performed on the character portion without performing image area separation, and both the image quality of the character portion, halftone dot portion, and flat portion can be achieved.
In the above description, the edge direction characteristic determination unit 12 determines whether the two edge amounts c and d in the diagonal direction are the maximum and the second largest value, and only when the two values are such a large value. The edge level conversion unit 13 multiplies α by 1; however, in order to carry out more precise processing, after referring to the edge direction information of neighboring peripheral pixels as a whole, Depending on the result of the directionality, the first edge level conversion unit 13 is configured to process various patterns, so that not only a 45 degree screen angle halftone image but also more types of directionality can be obtained. It is possible to control the enhancement / attenuation of the held edge information.
[0013]
【The invention's effect】
As described above, according to the present invention, together with the edge amount is detected from the input image data of multi-gradation, a first conversion corresponding to the frequency is performed on the input image data, and wherein The input image data is subjected to a second conversion according to the frequency, which is different from the first conversion, and the output result of the second conversion is multiplied according to the output result of the edge amount detection means. The output result of the multiplication and the output result of the first transformation are added, and in this case, in the first transformation, only the data component in a specific frequency domain is enhanced. High-quality edge images can be obtained by using multiplication output, which is an edge enhancement system output, for character portions and line portions other than character portions, and specific frequency enhancement is applied to halftone dots and flat portions. First conversion output that is a system output The output value can be obtained while suppressing moire and noise, and the image area discriminating means such as the image area separating means is not used, so the hardware scale does not increase and the cost does not increase. Just do it. In addition, both edge enhancement and specific frequency enhancement systems can show enhancement characteristics in the medium frequency range. Therefore, unnatural image quality degradation occurs at the switching between halftone dots, characters, and flat parts. Can also be suppressed .
Further, in the first conversion, a signal having a frequency transfer characteristic in an oblique 45-degree direction on a two-dimensional space showing the spatial frequency transfer characteristic is attenuated, so that moire in a halftone portion can be suppressed.
[0014]
Further, a plurality of directions of edge of the image plane from the input image data of multi-gradation, i.e. vertical, horizontal, left upward slanting edge of the four directions of the left-to-right upward slanting is detected, the edge of each detected direction intercomparison of is performed, since the converted edge amount for the pixel is calculated according to the comparison result, it detects a halftone portion, it is possible to reduce the edge of the halftone dot portion, therefore, or falling edge of Di-enhanced halftone dot edge amount can be suppressed.
Also, during or falling edge of di-amount conversion, as a result of comparison of edge amount of each direction detected, when the two diagonal directions is determined to be both strong edge direction characteristics, is a constant of 1 or less, it In other cases, since 1 is multiplied, it is possible to easily realize edge amount suppression for the halftone dot portion of the edge enhancement system .
Further, when a post-conversion edge amount, which is an output value of the first edge amount conversion, is input, a second edge amount conversion in which the output edge amount is saturated for a post-conversion edge amount equal to or greater than a predetermined amount. since performed, it is possible to suppress the edge amount output within a predetermined amount, thus, Ru prevents the output result is swung by abnormally large edge amount.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus in which the present invention is implemented.
FIG. 2 is a configuration block diagram of a main part of an image processing apparatus showing an example of an embodiment of the present invention.
FIG. 3 is an explanatory diagram of a main part of an image processing apparatus showing an example of an embodiment of the present invention.
FIG. 4 is another explanatory diagram of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
FIG. 5 is a block diagram of a main part of an image processing apparatus showing an example of an embodiment of the present invention.
FIG. 6 is another explanatory diagram of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
FIG. 7 is a processing flowchart of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
FIG. 8 is another explanatory diagram of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
FIG. 9 is another explanatory diagram of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
FIG. 10 is another explanatory diagram of the main part of the image processing apparatus showing an example of the embodiment of the present invention.
[Explanation of symbols]
1: image input unit 2: filter processing unit 11: edge amount detection unit 12: edge direction characteristic determination unit 13: first edge conversion unit 14: second edge conversion unit 15: first filter processing unit 16: first 2 filter processing unit 17: multiplier 18: adder

Claims (2)

In an image processing apparatus capable of performing conversion according to frequency on input image data by a spatial filter,
An edge amount detection means for detecting image plane a plurality of directions of edge amount of each of multi-tone input image data,
Edge direction characteristic determination means for comparing the edge amounts in each direction detected by the edge amount detection means;
If the comparison result of the edge direction characteristic determination means determines that both diagonal directions are strong edge direction characteristics, a constant of 1 or less is multiplied by 1 in other cases. First edge amount conversion means for calculating an edge amount;
Second edge amount conversion means having a non-linear conversion characteristic in which the output edge amount is saturated with respect to a post-conversion edge amount equal to or greater than a predetermined amount output from the first edge amount conversion unit;
Consists of a bandpass filter that extracts data components in a specific frequency domain and corrects the frequency transfer characteristics to attenuate signals with a frequency transfer characteristic of 45 degrees obliquely in a two-dimensional space showing the spatial frequency transfer characteristics and a first filtering means for applying a conversion corresponding to the frequency relative to the input image data,
And second filtering means for applying different, conversion corresponding to the frequency from said first filtering means to the input image data,
A multiplier for multiplying the output result of the second edge amount conversion means according to the output result and the second filtering means,
The image processing apparatus being characterized in that example Bei an adder for adding the output result of the output results from the first filtering means of the multiplier. In an image processing method capable of performing conversion according to frequency on input image data by a spatial filter,
  An edge amount detection step for detecting edge amounts in a plurality of directions on the image plane from multi-tone input image data;
  Edge direction characteristic determination step for comparing edge amounts in each direction detected by the edge amount detection step;
  If the comparison result of the edge direction characteristic determination step determines that the two diagonal directions are both strong edge direction characteristics, the result is converted by multiplying by a constant equal to or less than 1 otherwise. A first edge amount conversion step for calculating an edge amount;
  A second edge amount conversion step for performing non-linear conversion in which the output edge amount is saturated with respect to a post-conversion edge amount equal to or greater than a predetermined amount output from the first edge amount conversion step;
  Bandpass filtering is performed to extract data components in a specific frequency range and correct the frequency transfer characteristics so as to attenuate signals having a frequency transfer characteristic of 45 degrees obliquely in a two-dimensional space showing the spatial frequency transfer characteristics. A first filter processing step for performing conversion according to frequency on the input image data;
  A second filter processing step for performing conversion according to frequency, different from the first filter processing step, on the input image data;
  A multiplication step of multiplying the output result of the second edge amount conversion step by the output result of the second filter processing step;
  An image processing method comprising: an addition step of adding the output result of the multiplication step and the output result of the first filter processing step.

Images