JPH11328357A - Device and method for image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a binary image of good quality which is not noisy while eliminating texture characteristic of an error diffusing method when a multi- valued image is binarized. SOLUTION: By this image processing method, a noise generation part 206 generates noises of -5 to +5 in amplitude which are equal in occurrence probability, a threshold value generation part 207 adds an intermediate value '128' of an input density value range to the noise values to obtain a threshold value, and a data comparison part 205 binarizes a multi-valued image from an error adding process part 204 by using the threshold value. An error generation part 209 extracts error components at the time of the binarization, a weight adding process part 202 performs a weighted adding process and stores the result in an error memory, and an error adding process part 204 adds the error components to the input multi-valued image data to perform error diffusion.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and method, for example, an image processing apparatus and method for converting a multi-value (multi-tone) image into n-tones lower than the number of tones of an input image. It is.

[0002]

2. Description of the Related Art Conventionally, in order to print and output a multi-tone (multi-value) image from a binary printer, it is necessary to binarize the multi-value image by gradation conversion. As a method of binarizing a multi-valued image, there are methods such as a simple binarization method, a dither method, an error diffusion method, and an average density method.A combination of various methods based on these methods, Many gradation conversion methods have been devised and put into practical use.

[0003] In particular, when a multi-tone image such as a photographic image is output to a binary printer, a method based on the error diffusion method and the average density method can provide good tone expression. It is heavily used. Further, the error diffusion method and the average density method have a drawback that a specific texture depending on the method appears in the image after gradation conversion, but by adding random noise to the input multi-valued image, it is apparently added. This problem is also being resolved.

[0004] In the case of the error diffusion method,
In the region where the density of the input multi-valued image is exactly intermediate (2
In a 56-tone image, an area having a density of 128), a uniform texture appears as a result of the binarization processing, and this may appear as an apparent image contour (pseudo contour).

In the case of the average density method, black or white dots are often continuous for two or more pixels due to the algorithmic characteristics. In the method of adding random noise, the appearance of dots appearing at the time of binarization becomes random in order to add noise of a minute amplitude to an input image in advance. This makes it difficult for the specific arrangement of dots to appear as texture.

[0006] Pseudo contours caused by such a unique texture can be eliminated by adding random noise to an input image. However, on the other hand, noise is added to the image, so that the image after binarization becomes noisy.

[0007]

As described in the above-mentioned prior art, the method of adding random noise has an effect of alleviating a unique texture generated by an error diffusion method or the like. However, since noise addition for eliminating the texture is performed directly on the input image, there is a problem that the binarized image becomes noisy.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems. For example, when binarizing a multi-valued image, a unique texture such as an error diffusion method is eliminated, and at the same time, a good image without noise is obtained. The purpose is to obtain a binary image of high image quality.

[0009]

To achieve the above object, the present invention has, for example, the following constitution.

That is, an image processing apparatus for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein noise having the same occurrence probability is generated, and a predetermined value is added to the generated noise value And a threshold value generating means for generating a threshold value by adding the threshold value to the multi-valued image, and using the threshold value generated by the threshold value generating means to reduce the number of gradations to n gradations. Conversion means for conversion.

Further, for example, an error amount generating means for generating a difference value between the conversion value of the conversion means and the multi-valued pixel data value as an error amount, and an error amount generated by the error amount generating means is weighted by a predetermined weight. An error adding means is provided which accumulates and adds based on the coefficients, adds the multi-valued image data to the input, and converts the multi-valued image data into multi-valued image data for gradation conversion.

An image processing apparatus for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein the noise generation means generates noise having an equal probability of occurrence; The threshold value generating means for generating a predetermined threshold value based on the generated noise is compared with the threshold value generated by the threshold value generating means and the multi-valued image data for gradation conversion. A conversion unit for converting the multi-valued image data into n gradations having a smaller number of gradations than the multi-valued image data, and generating a difference value between the conversion value of the conversion unit and the multi-valued pixel data value as an error amount Error amount generating means, and multi-valued image data to be subjected to gradation conversion by accumulating and adding the error amount generated by the error amount generating means to input multi-valued image data based on a predetermined weighting coefficient. Error adding means. .

For example, the noise generation means generates noise having a predetermined amplitude, and the predetermined amplitude is sufficiently smaller than a range of the input multi-valued image data.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention according to the present invention will be described in detail with reference to the drawings.

FIG. 1 is a diagram showing an example in which a gradation conversion method in an image processing apparatus according to an embodiment of the present invention is realized as a means for binarizing a multi-valued image. Hereinafter, FIG.
The gradation conversion method in the present embodiment will be described with reference to FIG. The image processing apparatus shown in FIG. 1 is implemented as a digital image processing device. As digital image processing equipment, a copying machine, a facsimile and the like can be mentioned as typical examples.

In FIG. 1, reference numeral 101 denotes an image input means of a digital image processing device. Specific examples include an image reading device (scanner) using a CCD or a contact sensor, a digital camera, and the like. Further, an interface for inputting image data that has already been captured on a computer as digital image data may be used. The image input from the image input means is, for example, a 256 gradation multi-valued monochrome image of 8 bits per pixel.

Reference numeral 102 denotes an image processing means for performing predetermined image processing on the multi-valued image input by the image input means 101. The image processing performed by the image processing unit 102 has a function of converting luminance data into density data, a gamma correction function of correcting density according to density characteristics of a reading system and a recording system, and the like. Further, it has a filtering function for enhancing an edge portion of an image and removing moire or the like caused by an aliasing error generated in the course of digital image processing.

Reference numeral 103 denotes a binarizing unit which executes a gradation conversion method of a multi-valued image, which is a gradation conversion processing unit in the embodiment. In the present embodiment, a case is described in which a 256-tone image of 8 bits per pixel is converted into a 2-tone image of 1 bit per pixel. Therefore, the image data output from the binarization unit 103 and input to the image output unit 104 is
This is image data of one bit per pixel.

Reference numeral 104 denotes an image output unit, which corresponds to, for example, a binary printer that expresses digital data "0" and "1" in white and black, respectively. Note that the image output means of the present embodiment may be a general-purpose computer which simply takes in binarized image data, or in the case of a facsimile apparatus, may be a modem which modulates an image to a frequency required for communication.

Next, the binarizing means 103 of the embodiment shown in FIG. 1 will be described in more detail with reference to FIG. FIG. 2 is a diagram for explaining a detailed configuration of the binarizing means 103 which is the gradation conversion processing means shown in FIG.

In FIG. 2, reference numeral 201 denotes a pixel value input unit. As described above, one pixel is represented by 8 bits and has a value in a range from “0” to “255”. The pixel value input from the pixel value input unit 201 has been subjected to luminance density conversion, and “0” represents white and “255” represents black.

A weight addition processing unit 202 adds and accumulates the density error values generated up to the previous pixel in consideration of the weight coefficient. 203 is an error memory. The error memory 203 is used at the time of weight addition processing in the weight addition processing unit 202.

An error addition processing unit 204 adds the error amount generated in the error memory 203 by the weight addition processing unit 202 and the pixel input value of the pixel value input unit 201 on the preliminary route. Reference numeral 206 denotes a noise generation unit that generates digital noise. The noise generated by the noise generation unit 206 is random noise (white noise) whose range is limited in advance. The noise range generated by the noise generation unit 206 is −
5 to +5, but this value can be set arbitrarily.

Reference numeral 207 denotes a threshold generator. Conventionally, in the error diffusion method, the threshold value corresponding to this portion takes an intermediate value of the input density range and is fixed. However, in the present embodiment, this threshold value changes according to the output value of the noise generation unit 206 described above.

More specifically, the random noise in the range of -5 to +5 generated by the noise generator 206 is sent to the threshold generator 207 and a predetermined fixed value "12" is set.
8 ". As a result, a threshold value in the range of 123 to 133 is randomly generated as an output threshold value output from the threshold value generation unit 207 to the data comparison unit 205. Fixed value "128" is an intermediate value of the input density value range.

The noise randomly generated by the noise generating unit 206 is output in the range of -5 to +5 with the same probability of occurrence, so that macroscopically the average value of the threshold is "1".
28 ", so that the input pixel density can be preserved in the same manner as in the conventional error diffusion method. On the other hand, the threshold value takes a random value in the middle of the input density range, and the conventional texture is conspicuous. In the vicinity of the density value “128”, the texture can be selectively erased.

Reference numeral 205 shown in FIG. 2 is a data comparison unit.
The error addition value (pixel density value) output from the error addition processing unit 204 is compared with the threshold value output from the threshold value generation unit 207. If the pixel density value is equal to or greater than the threshold value, the data comparison means 205 outputs "1". Conversely, if the pixel density value is less than the threshold value, the data comparison means 205 outputs “0”.

The output value from the data comparison unit 205 is
This is binary data in which one pixel is represented by one bit of “0” and “1”.

In the present embodiment described above, binarization processing of a multi-valued image has been described as gradation conversion processing. The output can be multi-valued.

For example, when there are three threshold values, the data comparison unit 205 can output four types of results, and the output image in this case is a 2-bit quaternary output. Accordingly, not only the above-described multi-value / binary conversion, but also gradation conversion such as multi-value / multi-value conversion can be realized with exactly the same configuration.

A data output unit 208 in FIG. 2 outputs image data binarized by the data comparison unit 205. The binarized image data is transferred to the image output unit 104 shown in FIG. 1 to obtain a predetermined output image.

Reference numeral 209 in FIG. 2 denotes an error generating unit. Since the binarized image data has only two gradations of "1" or "0", a deviation from the actual density value occurs. Therefore, in the present embodiment, the error generation unit 209 generates a deviation from the actual density value as a density error value based on an algorithm of a normal error diffusion method, and generates a deviation in a subsequent pixel binarization process. This is taken into account.

That is, when the output image is "1", a value obtained by subtracting "255" from the input multi-valued image data is output as an error amount. On the other hand, when the output image is "0", the input image data is left as it is. Output as error amount.

Then, as described above, the weight addition processing unit 201 and the error memory 203 multiply and add the error generated by the error generation unit 209 in consideration of a predetermined weight coefficient.
The result is stored in the error memory 203.

In the weight addition processing in the weight addition processing unit 202, the generated error is divided and diffused into a predetermined two-dimensional area centered on the pixel of interest in accordance with the weighting coefficient. There is a method (minimum error method) of accumulating data in a predetermined two-dimensional area in consideration of coefficients.
In this embodiment, any of the methods can be applied.

As described above, the multi-valued input image is subjected to the binarization processing, and can be output to the binary printer. Conventionally,
In the error diffusion method, the threshold value for binarization is fixed and does not change. This is because binarization is performed while maintaining the density of the input multi-valued image. Therefore, if noise is added to this threshold value, the density will not be preserved in the character. However, by making the occurrence probabilities of all the values of the additional noise the same, the expected value of the threshold to be generated can be made constant. For this reason, in the present embodiment, binarization of an image can be performed macroscopically while keeping the density of the input image.

Further, since noise is not directly added to image data as in the conventional example but is added to a threshold value for binarization and noise is added indirectly and selectively with respect to density, image noise is reduced. This alleviates the problem of noisy output as a whole. Further, the texture peculiar to the error diffusion method can be eliminated by selectively adding noise.

As described above, according to the present embodiment, random noise is a threshold value for executing binarization without directly adding noise to input multi-valued image data as in the conventional method. Is added to the output binary image, random noise for eliminating the texture unique to the error diffusion method is indirectly superimposed via the threshold value.

For this reason, since the noise power can be limitedly added to the vicinity of the threshold value, a noisy image over the entire density region, which is a problem in the conventional method, is alleviated, and a relatively good image quality is obtained. A binary image can be obtained.

Further, the unique texture which has been conspicuously appearing in the middle portion of the input image density range can be selectively erased by randomly changing the threshold value.

[Other Embodiments] Even if the present invention is applied to a system including a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), an apparatus including a single device can be used. (For example, a copying machine, a facsimile machine, etc.).

Another object of the present invention is to provide a storage medium storing a program code of software for realizing the functions of the above-described embodiments to a system or apparatus, and to provide a computer (or CPU) of the system or apparatus.
And MPU) read and execute the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

As a storage medium for supplying the program code, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD
-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

When the computer executes the readout program code, not only the functions of the above-described embodiment are realized, but also the OS (Operating System) running on the computer based on the instruction of the program code. ) May perform some or all of the actual processing, and the processing may realize the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, based on the instructions of the program code, It goes without saying that the CPU included in the function expansion board or the function expansion unit performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

When the present invention is applied to the storage medium, the storage medium stores program codes for implementing the functions described above.

[0048]

As described above, according to the present invention, the texture peculiar to the error diffusion method can be eliminated, and the image quality can be improved by eliminating the false contour appearing in the smooth gradation image portion. On the other hand, a noisy image due to the side effect of the additional noise as in the conventional example is mitigated by the indirect addition of the noise to the threshold value, and the effect of further improving the image quality is obtained.

[0049]

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an example in which a gradation conversion method in an image processing apparatus according to an embodiment of the present invention is implemented as a binarizing unit of a multi-value image.

FIG. 2 is a diagram for explaining a detailed configuration of a binarizing unit which is a gradation conversion processing unit shown in FIG. 1;

[Explanation of symbols]

 Reference Signs List 101 image input means 102 image processing means 103 binarization means 104 image output means 201 pixel value input section 202 weight addition processing section 203 error memory 204 error addition processing section 205 data comparison section 206 noise generation section 207 threshold value generation section 208 Data output unit 209 Error generation unit

Claims (10)

[Claims] 1. An image processing apparatus for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein noise having an equal probability of occurrence is generated, and a predetermined value is set to the generated noise value. Threshold value generating means for generating a threshold value by adding the threshold value to the multi-valued image, and using the threshold value generated by the threshold value generating means to n gradations having a smaller number of gradations than the multi-valued image. An image processing apparatus comprising: a conversion unit configured to perform conversion. 2. An error amount generating means for generating a difference value between a converted value of the converting means and a multi-valued pixel data value as an error amount; 2. The image processing apparatus according to claim 1, further comprising: an error adding unit that accumulates and adds to the input multi-valued image data based on the data and adds the multi-valued image data to multi-valued image data for gradation conversion. 3. An image processing apparatus for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein: a noise generating means for generating noise having an equal probability of occurrence; Threshold value generating means for generating a predetermined threshold value based on the generated noise; comparing the threshold value generated by the threshold value generating means with multi-valued image data for gradation conversion; A conversion means for converting the multi-valued image data into n gradations having a smaller number of gradations than the multi-valued image data; and generating a difference value between the conversion value of the conversion means and the multi-valued pixel data value as an error amount. An error amount generating unit that performs accumulation and addition of the error amount generated by the error amount generating unit on the basis of a predetermined weighting coefficient to input multi-valued image data to obtain multi-valued image data for gradation conversion Error adding means is provided. An image processing device. 4. The apparatus according to claim 1, wherein said noise generating means generates noise having a predetermined amplitude, and said predetermined amplitude is sufficiently smaller than a range of input multi-valued image data. The image processing device according to any one of the above. 5. An image processing method for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein noise having an equal probability of occurrence is generated, and a predetermined value is added to the generated noise value. An image processing method comprising: converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image using the generated threshold value. 6. The method according to claim 1, wherein a difference value between the converted value and the multi-valued pixel data value is used as an error amount, and the error amount is accumulated and added based on a predetermined weighting coefficient and added to the input multi-valued image data. 6. The image processing method according to claim 5, wherein the image data is multi-valued image data to be subjected to gradation conversion. 7. An image processing method for converting a multi-valued image into n gradations having a smaller number of gradations than the multi-valued image, wherein noise having an equal probability of occurrence is generated, and a predetermined value is determined based on the generated noise. A threshold value is generated, the generated threshold value is compared with the multi-valued image data to be subjected to gradation conversion, and the multi-valued image data is converted into n-order images having a smaller number of gradations than the multi-valued image data according to the comparison result And converting the difference value between the converted value and the multi-valued pixel data value into an error amount, and accumulating and adding the error amount based on a predetermined weight coefficient to the input multi-valued image data. An image processing method characterized by using multi-valued image data for gradation conversion. 8. The method according to claim 5, wherein the generated noise is noise having a predetermined amplitude, and the predetermined amplitude is sufficiently smaller than a range of input multi-valued image data. The image processing method according to any one of the above. 9. A computer program sequence for realizing the function according to any one of claims 1 to 8. 10. A computer-readable recording medium storing a computer program for realizing the functions according to claim 1. Description: