JP4293255B2 - Image processing method and image processing apparatus - Google Patents

Description

  The present invention relates to an image processing method and apparatus for performing gradation reduction of image data by performing error diffusion processing.

  For example, when a multi-gradation image created by a computer or input by an image input device is output by an image output device such as a printer with fewer gradations, Need to be done. As described above, pseudo halftone expression has been used as a technique for maintaining the image quality of the original image as much as possible even when the gradation of the image data is reduced. Various methods of pseudo-halftone expression have been proposed. Among them, pseudo-halftone expression by the error diffusion method is widely used in printers having a binary output gradation because of good image quality. The error diffusion method is a method of diffusing the quantization error generated for the target pixel to the input image data of the unquantized pixel in the vicinity of the target pixel.

  Here, the principle of a general error diffusion method (see Non-Patent Document 1) will be described in detail.

  The error diffusion method is a method of expressing a pseudo halftone by making the quantization error inconspicuous by taking human visual characteristics into account and modulating it. FIG. 10 is a block diagram showing a configuration of an image processing apparatus for realizing general error diffusion processing. This image processing apparatus includes a subtractor 111 that subtracts output data of a filter 114 (to be described later) from input image data x (i, j), and quantizes the output data of the subtractor 111 to output image data y (i, j), a quantizer (denoted as Q in the figure) 112, a subtractor 113 for subtracting the output data of the subtractor 111 from the output image data y (i, j), and output data of the subtractor 113 And a filter 114 that performs a predetermined filtering process and outputs the result to the subtractor 111. In the figure, e (i, j) represents a quantization error caused by quantization in the quantizer 112. Therefore, the output data of the subtracter 113 becomes a quantization error e (i, j). Note that i and j represent coordinates in two directions orthogonal to each other (hereinafter referred to as i direction and j direction).

The filter 114 is a kind of linear filter, and here, its transfer function is G (z _1, z ₂ ). Note that z _{1 and} z ₂ are variables in z-conversion regarding the i direction and the j direction, respectively. The entire configuration of the image processing apparatus shown in FIG. 10 can be regarded as a two-dimensional ΣΔ modulation circuit. Therefore, the input / output relationship in this image processing apparatus is given by the following equation (1).

[Equation 1]
_{Y (z 1, z 2)} = X (z 1, z 2) + H (z 1, z 2) E (z 1, z 2) ... (1)

In equation (1), Y (z _1, z ₂ ), X (z _1, z ₂ ), and E (z _1, z ₂ ) are y (i, j) and x (i, j), respectively. ), E (i, j) are values obtained by z conversion. The transfer function H (z _1, z ₂ ) of the filter that modulates the quantization error E (z _1, z ₂ ) is given by the following equation (2).

[Equation 2]
_{H (z 1, z 2)} = 1-G (z 1, z 2) ... (2)

The transfer function H (z _1, z ₂ ) represents a two-dimensional finite impulse response (FIR) high pass filter, which modulates the quantization error E (z _1, z ₂ ) to the high range. It becomes an error modulation filter that determines the characteristics. In the following description, the filter H (z _1, z ₂ ) and the filter G (z _1, z ₁ ) are also used to represent the filters represented by the transfer functions H (z _1, z ₂ ) and G (z _1, z ₂ ) _. z ₂ ).

G (z _1, z ₂ ) is expressed as the following formula (3).

[Equation 3]
_{G (z 1, z 2)} = ΣΣg (n1, n2) z 1 -n1 z 2 -n2 ... (3)

Note that the first Σ in equation (3) represents the sum of n1 from −N ₁ to M ₁ , and the next Σ represents the sum of n2 from −N ₂ to M ₂ . Here, N ₁ , M ₁ , N ₂ , and M ₂ are predetermined positive integers. Further, g (n1, n2) is a filter coefficient, and n1 = 0 and n2 = 0 represent a target pixel.

Here, an example of a typical filter is given as an example of the coefficient g (i, j) of G (z _1, z ₂ ) in Expression (4). Note that * in the expression represents a target pixel, and g (0, 0) = 0.

FIG. 11 shows the frequency characteristics of the error modulation filter H (z _1, z ₂ ) using the filter G (z _1, z ₂ ) represented by Expression (4). In FIG. 11, the numerical value representing the frequency represents that the frequency is higher as the absolute value is larger. The filter G (z _1, z ₂ ) represented by the expression (4) and the filter H (z _1, z ₂ ) using the filter G (z _1, z ₂ ) are called Jarvis, Judice & Ninke filters (hereinafter referred to as Jarvis filters). Is.

Hitoshi Kiya, and one other, “Tone Conversion Method for Digital Image Data Using C”, Interface, August 1993, p. 158-171

  However, in the conventional error diffusion method as described above, dot dispersibility is poor in the highlight area (the area where the dots are sparse) and the shadow area (the area where the dots are dense), and the error diffusion method has a unique “worm”. There was a problem that a pattern called this occurred and this deteriorated the image quality.

  FIG. 12 shows an example of an image with poor dot dispersibility and a “worm” pattern. This figure shows an image obtained by subjecting an image whose gradation value gradually changes in the vertical direction (hereinafter referred to as a vertical ramp image) to error diffusion processing using the above-mentioned general Jarvis filter. The highlighted part is shown. In the image shown in this figure, the dots are not uniformly distributed, but a “worm” pattern, which is a series of dots like insects, appears, which deteriorates the image quality.

  Here, the reason why the dot dispersibility deteriorates in the conventional error diffusion method will be considered from the viewpoint of the frequency characteristics of the error modulation filter. FIG. 13 is a diagram expressing the frequency characteristics of the Jarvis filter shown in FIG. 11 using contour lines. With such an expression, the spatial shape (characteristic) of the frequency characteristic of the filter is well understood. As can be seen from FIG. 13, since the coefficient of the error modulation filter is spatially asymmetric as shown in Equation (4), the frequency characteristics of the filter are also spatially asymmetric. In particular, the spatial distortion is large in the central portion of the frequency characteristic shown in FIG. 13, that is, the characteristic portion of the quantization error with respect to the low frequency region. The dot dispersibility becomes a problem because there are many flat parts such as highlight areas and shadow areas. Therefore, the frequency characteristics of the output image data are caused by the spatial distortion at the center of the frequency characteristics shown in FIG. It is considered that a spatial deviation occurs in the image, which deteriorates the dispersibility of the dots and produces a “worm” pattern.

  Accordingly, it is considered that the problem of dot dispersibility can be solved by improving the characteristics of the error modulation filter with respect to the low frequency region so as to be more spatially symmetric.

  However, if an attempt is made to improve the characteristics of the error modulation filter in the low frequency range, the distortion in the peripheral portion of the frequency characteristics shown in FIG. "Worm" pattern is generated. As described above, it is difficult for the error modulation filter to realize spatially symmetric characteristics in all frequency bands.

  The present invention has been made in view of such problems, and an object of the present invention is to provide an image processing method capable of preventing image quality deterioration while performing error diffusion processing and reducing the gradation of image data. To provide an apparatus.

The image processing method of the present invention includes a quantization procedure for quantizing input image data for each pixel and converting the input image data into output image data having one of two or more quantization levels, and input image data of a target pixel. The weighting factor selected by the weighting factor selection procedure is added to the weighting factor selection procedure for selecting a weighting factor for error diffusion according to the gradation value, and the quantization error generated by the quantization in the quantization procedure for the target pixel. And an error diffusion procedure for diffusing the input image data of unquantized pixels in the vicinity of the pixel of interest by multiplication. In the weighting factor selection procedure, the gradation value of the input image data of the target pixel is within the first range including the values farthest from the minimum value and the maximum value in the gradation value range of the input image data. Is included, the first weighting factor composed of a plurality of elements is selected, and the gradation value of the input image data of the target pixel includes the minimum value or the maximum value in the range of gradation values of the input image data. When included in the predetermined second range, a second weighting factor composed of a plurality of elements including elements different from the elements in the first weighting coefficient is selected, and the gradation value of the input image data of the target pixel is When included in the range between the first range and the second range, each element has an element obtained by linear interpolation based on the first weighting coefficient and the second weighting coefficient. A weighting factor is selected.

An image processing apparatus according to the present invention includes a quantization unit that quantizes input image data for each pixel and converts the input image data into output image data having one of two or more quantization levels, and input image data of a target pixel. The weighting coefficient selection means for selecting a weighting coefficient for error diffusion according to the gradation value, and the weighting coefficient selected by the weighting coefficient selection means for the quantization error generated by the quantization in the quantization means for the target pixel. And an error diffusing unit that multiplies and diffuses the input image data of unquantized pixels in the vicinity of the target pixel. Further, the weighting factor selecting means has a gradation value of the input image data of the target pixel within a first range including a value farthest from a minimum value and a maximum value in a range of gradation values of the input image data. Is included, the first weighting factor composed of a plurality of elements is selected, and the gradation value of the input image data of the target pixel includes the minimum value or the maximum value in the range of gradation values of the input image data. When included in the predetermined second range, a second weighting factor composed of a plurality of elements including elements different from the elements in the first weighting coefficient is selected, and the gradation value of the input image data of the target pixel is When included in the range between the first range and the second range, each element has an element obtained by linear interpolation based on the first weighting coefficient and the second weighting coefficient. A weighting factor is selected.

In the image processing method of the present invention, the input image data is quantized for each pixel by the quantization procedure, and converted into output image data having one of two or more quantization levels. Further, the weighting factor for error diffusion is selected according to the gradation value of the input image data of the target pixel by the weighting factor selection procedure, and the quantum generated by the quantization in the quantization procedure is performed for the target pixel by the error diffusion procedure. The quantization error is multiplied by the weighting factor selected by the weighting factor selection procedure, and is diffused to the input image data of the unquantized pixel in the vicinity of the target pixel. In addition, when the weighting factor is selected, the gradation value of the input image data of the target pixel includes the first value that is farthest from the minimum value and the maximum value in the gradation value range of the input image data. Is included, the first weighting factor composed of a plurality of elements is selected, and the gradation value of the input image data of the target pixel is the minimum value or the maximum value in the gradation value range of the input image data. When included in a predetermined second range including a value, a second weighting factor composed of a plurality of elements including elements different from the elements in the first weighting coefficient is selected, and the scale of the input image data of the target pixel is selected. When the tone value is included in the range between the first range and the second range, it is obtained by linear interpolation for each element based on the first weighting factor and the second weighting factor. A weighting factor with elements is selected.

In the image processing apparatus of the present invention, the input image data is quantized for each pixel by the quantization means and converted into output image data having one of two or more quantization levels. In addition, the weighting factor selection unit selects a weighting factor for error diffusion according to the gradation value of the input image data of the target pixel, and the error diffusion unit causes the target pixel to be generated by quantization in the quantization unit. The quantization error is multiplied by the weighting factor selected by the weighting factor selection means, and is diffused to the input image data of the unquantized pixel in the vicinity of the target pixel. In addition, when the weighting factor is selected, the gradation value of the input image data of the target pixel includes the first value that is farthest from the minimum value and the maximum value in the gradation value range of the input image data. Is included, the first weighting factor composed of a plurality of elements is selected, and the gradation value of the input image data of the target pixel is the minimum value or the maximum value in the gradation value range of the input image data. When included in a predetermined second range including a value, a second weighting factor composed of a plurality of elements including elements different from the elements in the first weighting coefficient is selected, and the scale of the input image data of the target pixel is selected. When the tone value is included in the range between the first range and the second range, it is obtained by linear interpolation for each element based on the first weighting factor and the second weighting factor. A weighting factor with elements is selected.

According to the image processing method or the image processing apparatus of the present invention, the input image data is quantized for each pixel and converted into output image data having one of two or more quantization levels, and the input of the target pixel is performed. Select the weighting factor for error diffusion according to the tone value of the image data, multiply the quantization error generated by quantization for the pixel of interest by the selected weighting factor, and unquantize in the vicinity of the pixel of interest Since the pixel input image data is diffused, an error diffusion process is performed to reduce the gradation of the image data, and the image quality can be prevented from being deteriorated.
Further, when selecting the weighting factor, the first range in which the gradation value of the input image data of the target pixel includes a value farthest from the minimum value and the maximum value in the range of the gradation value of the input image data. A first weighting factor composed of a plurality of elements is selected, and the gradation value of the input image data of the target pixel is set to a minimum value or a maximum value in a range of gradation values of the input image data. A second weighting factor composed of a plurality of elements including elements different from the elements in the first weighting factor is selected, and the gradation value of the input image data of the target pixel is selected. Is included in the range between the first range and the second range, the elements obtained by linear interpolation for each element based on the first weighting factor and the second weighting factor are Since the weighting factor is selected, discontinuous switching No image deterioration due an effect that it is possible to obtain an output image data with smooth gradation.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of an image processing apparatus for realizing an image processing method according to the first embodiment of the present invention. The image processing apparatus 10 includes a subtractor 11 that subtracts output data of a filter unit 14 (to be described later) from input image data x (i, j), and outputs the output data of the subtracter 11 to one or more pixels for each pixel. A quantizer (denoted as Q in the figure) 12 as a quantization means for performing quantization based on a threshold value and converting it into output image data y (i, j) having one of two or more quantization levels; The subtracter 13 that subtracts the output data of the subtractor 11 from the output data of the quantizer 12 to generate a quantization error e (i, j), and the output data of the subtractor 13 has a predetermined value. The filter coefficient in the filter unit 14 that performs filtering processing and outputs to the subtractor 11 and the filter unit 14 corresponding to the weight coefficient for error diffusion is determined according to the gradation value of the input image data x (i, j). Select and filter part 1 And a weighting coefficient selection unit 15 to be set to.

  The filter unit 14 and the subtractor 11 multiply the quantization error e (i, j) generated by the quantization in the quantizer 12 for the pixel of interest by the weight coefficient (filter coefficient) selected by the weight coefficient selection unit 15. This corresponds to error diffusion means for diffusing the input image data of unquantized pixels in the vicinity of the pixel of interest, and modulates the quantization error to a high frequency region with a modulation characteristic corresponding to the weighting factor.

The filter unit 14 constitutes a kind of linear filter, and its transfer function is G (z _1, z ₂ ) here. Note that z _{1 and} z ₂ are variables in z-conversion regarding the i direction and the j direction, respectively. G (z _1, z ₂ ) is expressed as in the above equation (3). The filter unit 14 is realized by a digital filter, for example.

  The input image data x (i, j) input to the image processing apparatus 10 is given from, for example, the image input apparatus 1, and the output image data y (i, j) output from the image processing apparatus 10 is, for example, It is output to the image output device 2. Examples of the image input device 1 include an image scanner, a digital camera, and a video camera. Examples of the image output device 2 include a printer and a liquid crystal display.

  Note that the image processing apparatus 10 according to the present embodiment may be configured as, for example, a single apparatus, or may be incorporated in an image output apparatus such as a printer by being integrated into an IC (integrated circuit). Good.

  In the present embodiment, the gradation value of the input image data x (i, j) is set to 256 levels from “0” to “255”, and the final output image data y (i, j) is set to binary. In addition, each quantization level in the quantizer 12 is set to “0” and “255”.

  The weighting factor selection unit 15 indicates that the modulation characteristic of the quantization error becomes lower in frequency range as the gradation value of the input image data x (i, j) of the target pixel is closer to the minimum or maximum quantization level in the quantizer 12. The weighting coefficient is selected so as to be more symmetrical with respect to space.

  Specifically, the weight coefficient selection unit 15 determines that the tone value of the input image data x (i, j) of the target pixel is the minimum value “0” and the maximum value in the tone value range of the input image data. When included in the first range (12 <x (i, j) <243) that includes the value farthest from “255”, the first weighting coefficient composed of a plurality of elements is selected and the pixel of interest is input. The gradation value of the image data x (i, j) is a predetermined second range (x (i, j,) including the minimum value “0” or the maximum value “255” in the gradation value range of the input image data. j) ≦ 5, x (i, j) ≧ 250), a second weighting factor consisting of a plurality of elements including elements different from the elements in the first weighting factor is selected, and the pixel of interest When the gradation value of the input image data x (i, j) is included in a range between the first range and the second range In is adapted to, based on the first and second weighting coefficients, to select a weighting coefficient with a resulting element by linear interpolation for each element.

  Hereinafter, the filter realized by the filter unit 14 when the first weighting factor is selected is referred to as a halftone region filter A, and the filter realized by the filter unit 14 when the second weighting factor is selected. Is referred to as a highlight area filter B. An example of the filter coefficient g (i, j) of the filter A is shown in Expression (5), and an example of the filter coefficient g (i, j) of the filter B is shown in Expression (6).

FIG. 2 is a diagram in which the frequency characteristics of the error modulation filter H (z _1, z ₂ ) using the filter A represented by Expression (5) are expressed using contour lines, and FIG. FIG. 6 is a diagram expressing the frequency characteristics of an error modulation filter H (z _1, z ₂ ) using a filter B represented by () using contour lines.

  Referring to FIG. 2, in the halftone region filter A, the peripheral portion, that is, the characteristic portion of the quantization error with respect to the high frequency region is spatially almost symmetrical, and the spatial distortion of the texture is present in the halftone region. It can be seen that the filter does not occur. On the other hand, in FIG. 3, in the highlight filter B, the central portion, that is, the characteristic portion of the quantization error with respect to the low frequency region is spatially almost symmetric (circular), and the highlight region and the shadow region It can be seen that the filter has a good dot dispersibility and does not generate a “worm” pattern.

  The gradation value of the input image data x (i, j) of the target pixel is a range between the first range and the second range, that is, 6 ≦ x (i, j) ≦ 12 or 243 ≦ x ( i, j) ≦ 249, the filter realized by the filter unit 14 is a filter coefficient obtained by linear interpolation for each filter coefficient based on the filter coefficient of the filter A and the filter coefficient of the filter B. It becomes the filter which consists of. Thus, in the range between the first range and the second range, the filter coefficients are switched stepwise.

4 to 6 show frequency characteristics of the error modulation filter H (z _1, z ₂ ) using each filter when the filter coefficient is switched in stages from the filter coefficient of the filter A to the filter coefficient of the filter B. It is the figure which expressed using a contour line.

  FIG. 4A shows frequency characteristics corresponding to the filter A. FIG. 4B shows frequency characteristics corresponding to a filter selected when the gradation value of the input image data x (i, j) of the target pixel is “12” or “243”. When each filter coefficient of the filter A is represented by “A” and each filter coefficient of the filter B is represented by “B”, each filter coefficient corresponding to the frequency characteristic shown in FIG. The filter coefficient is (7A + B) / 8. FIG. 4C shows the frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “11” or “244”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 4C is (6A + 2B) / 8.

  FIG. 5A shows the frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “10” or “245”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 5A is (5A + 3B) / 8. FIG. 5B shows frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “9” or “246”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 5B is (4A + 4B) / 8. FIG. 5C shows the frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “8” or “247”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 5C is (3A + 5B) / 8.

  FIG. 6A shows the frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “7” or “248”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 6A is (2A + 6B) / 8. FIG. 6B shows frequency characteristics corresponding to the filter selected when the gradation value of the input image data x (i, j) of the target pixel is “6” or “249”. Each filter coefficient of the filter corresponding to the frequency characteristic shown in FIG. 6B is (A + 7B) / 8. FIG. 6C shows frequency characteristics corresponding to the filter B.

  Next, the operation of the image processing apparatus 10 according to the present embodiment will be described. The following description also serves as a description of the image processing method according to the present embodiment.

  In the image processing apparatus 10 according to the present embodiment, the subtracter 11 subtracts the output data of the filter unit 14 from the input image data x (i, j). The output data of the subtractor 11 is quantized by the quantizer 12, and the output data of the quantizer 12 is output from the image processing apparatus 10 as output image data y (i, j). Further, the subtracter 13 subtracts the output data of the subtracter 11 from the output data of the quantizer 12 to generate a quantization error e (i, j). The quantization error e (i, j) that is output data of the subtractor 13 is input to the filter unit 14 and subjected to a filtering process according to the filter coefficient (weight coefficient) selected by the weight coefficient selection unit 15. The The output data of the filter unit 14 is input to the subtractor 11. The weight coefficient selection unit 15 selects a filter coefficient in the filter unit 14 corresponding to the weight coefficient for error diffusion according to the gradation value of the input image data x (i, j), and sets it in the filter unit 14.

  With such an operation, the image processing apparatus 10 quantizes the input image data x (i, j) based on one threshold value, and outputs image data y (i, j, having one of two quantization levels. j), and the quantization error e (i, j) generated by this quantization is multiplied by the weighting factor selected by the weighting factor selection unit 15 to input an unquantized pixel in the vicinity of the target pixel. An error diffusion process for diffusing the image data x (i, j) is performed.

  In the image processing method and apparatus according to the present embodiment, the gradation value of the input image data x (i, j) of the target pixel has a first range (12 <x (i, j) < 243), the filter realized by the filter unit 14 is a filter A in which the characteristic portion of the quantization error with respect to the high frequency region is spatially symmetric, and the input image data x (i, j of the pixel of interest ) Is included in the second range (x (i, j) ≦ 5, x (i, j) ≧ 250) which is a highlight region or a shadow region, the filter unit 14 The realized filter is a filter B whose characteristic of the quantization error with respect to the low frequency region is spatially symmetric, and the gradation value of the input image data x (i, j) of the target pixel is in the first range. And the second range (6 ≦ x (i, j ≦ 12 or 243 ≦ x (i, j) ≦ 249), the filter realized by the filter unit 14 is determined for each filter coefficient based on the filter coefficient of the filter A and the filter coefficient of the filter B. And a filter composed of filter coefficients obtained by linear interpolation. As a result, according to the present embodiment, the dots are uniformly dispersed in the highlight area and the shadow area, and the dots are also uniformly dispersed in the halftone area. Therefore, according to the present embodiment, it is possible to prevent image quality deterioration while performing error diffusion processing to reduce the gradation of image data, and to obtain a high-quality image having almost no “worm” pattern. Can do.

  Further, according to the present embodiment, in the range between the first range and the second range, by the filter unit 14 according to the gradation value of the input image data x (i, j) of the target pixel. Since the realized filter is switched step by step between the filter A and the filter B, the frequency characteristics of the filter change smoothly, and there is no image quality deterioration due to the frequency characteristics of the filter being switched discontinuously. Thus, output image data having smooth gradation can be obtained.

FIG. 7 shows a highlight portion of an image obtained by performing error diffusion processing on the vertical lamp image by the image processing apparatus 10 according to the present embodiment. Comparing FIG. 7 and FIG. 12, it can be seen that according to the present embodiment, an output image in which dots are uniformly dispersed and an “worm” pattern is hardly generated can be obtained.

  In addition, the image processing method and apparatus according to the present embodiment can be realized by adding a very simple process (weight coefficient selection process) to a general error diffusion process. Compared with processing, processing time hardly increases.

  FIG. 8 is a block diagram showing a configuration of an image processing apparatus for realizing an image processing method according to the second embodiment of the present invention. The present embodiment is an example in which the average error minimum method is used instead of the error diffusion method. In the image processing apparatus 60 according to the present embodiment, the image processing apparatus 10 according to the first embodiment shown in FIG. 1 is provided with a filter unit 64 for the minimum average error method instead of the filter unit 14. Instead of the weighting factor selecting unit 15, a filtering factor in the filtering unit 64 is selected and a weighting factor selecting unit 65 for setting in the filtering unit 64 is provided. The filter unit 64 in the present embodiment holds the quantization error e (i, j) in a plurality of quantized pixels in the vicinity of the target pixel, and against these quantization errors e (i, j). Then, the average error is calculated by multiplying each by a predetermined weight coefficient, and the average error is output when the input image data of the target pixel is input. Such processing is equivalent to diffusing the quantization error generated for the target pixel with respect to the input image data of the unquantized pixel in the vicinity of the target pixel. The filter unit 64 can be realized by a digital filter, for example.

  Other configurations, operations, and effects in the present embodiment are the same as those in the first embodiment.

  FIG. 9 is a block diagram showing a configuration of an image processing apparatus according to the third embodiment of the present invention. The present embodiment is an example in which functions similar to those of the image processing apparatus 10 according to the first embodiment are realized by software using a computer.

  The image processing apparatus according to the present embodiment uses a computer, and includes a CPU (Central Processing Unit) 31, a ROM 32, and a RAM (Random Access Memory) 33 that are connected to each other via a bus 30. Yes. The image processing apparatus further includes a hard disk drive 51, a CD (compact disk) -ROM drive 52, a floppy disk drive 53, a keyboard 54, a mouse 55, and a CRT (cathode ray tube) connected to the bus 30 via interfaces 41 to 46. 56. The image processing apparatus further includes an interface 47 for connecting the image input device 57 to the bus 30 and an interface 48 for connecting the image output device 58 to the bus 30.

  Examples of the image input device 57 include an image scanner, a digital camera, and a video camera. Examples of the image output device 58 include a printer and a liquid crystal display.

  In the image processing apparatus according to the present embodiment, the CPU 31 uses the RAM 33 as a work area, the hard disk in the hard disk drive 51, the CD-ROM driven by the CD-ROM drive 52, or the floppy disk driven by the floppy disk drive 53. The functions of the image processing apparatus 10 in FIG. 1 are realized by executing an application program stored on the disk.

  The image processing apparatus according to the present embodiment performs the first operation on the image data input by the image input apparatus 57 or the image data created by the image processing apparatus (computer) by the function realized as described above. The same processing as in the embodiment is performed, and the output image data with reduced gradation is output to the image output device 58.

  Other operations and effects in the present embodiment are the same as those in the first embodiment.

  Note that the same functions as those of the image processing apparatus 60 according to the second embodiment may be realized by software using a computer, as in the third embodiment.

  Note that the present invention is not limited to the above-described embodiments. For example, the quantization level, the number of quantization levels, filter characteristics, and the like given in the embodiments are examples, and the present invention is applied to the embodiments. It can be set as appropriate.

  In the embodiment, the filter of the filter realized by the filter unit 14 when the gradation value of the input image data of the target pixel is included in the range between the first range and the second range. The coefficients are obtained by linear interpolation for each filter coefficient based on the filter coefficient of the filter A and the filter coefficient of the filter B. However, the filter unit 14 has the filter coefficient of the gradation value of the input image data. By using an adaptive filter as a function, the filter coefficient may be switched according to the gradation value of the input image data.

  In the embodiment, the input image data is quantized based on one threshold value and converted into output image data of two gradations. However, the present invention provides two or more input image data. The present invention can also be applied to the case where the output image data is quantized based on the threshold value and converted to output image data having three or more gradations.

  In addition, the present invention is effective for reducing the gradation of image data in order to output the image data to an inkjet printer, a fusion thermal transfer method or a thermo-autochrome printer, a display device with low gradation expression, and the like. However, it is also effective when reducing the gradation of image data in order to reduce the burden of image processing and image data storage.

1 is a block diagram illustrating a configuration of an image processing apparatus according to a first embodiment of the present invention. It is explanatory drawing which expressed the frequency characteristic of the filter A selected by the weighting coefficient selection part in FIG. 1 with the contour line. It is explanatory drawing which expressed the frequency characteristic of the filter B selected by the weighting coefficient selection part in FIG. 1 with the contour line. It is explanatory drawing which expressed the frequency characteristic of the filter obtained by linear interpolation based on the filter A and the filter B with the contour line. It is explanatory drawing which expressed the frequency characteristic of the filter obtained by linear interpolation based on the filter A and the filter B with the contour line. It is explanatory drawing which expressed the frequency characteristic of the filter obtained by linear interpolation based on the filter A and the filter B with the contour line. It is explanatory drawing which shows the image obtained by performing an error diffusion process with the image processing apparatus which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structure of the image processing apparatus which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the image processing apparatus which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the image processing apparatus for implement | achieving a general error diffusion process. It is a characteristic view which shows an example of the frequency characteristic of the filter used with the image processing apparatus shown in FIG. It is explanatory drawing which shows the image obtained by performing an error diffusion process with the image processing apparatus shown in FIG. It is explanatory drawing which expressed the frequency characteristic of the filter shown in FIG. 11 using the contour line.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Image processing apparatus, 11, 13 ... Subtractor, 12 ... Quantizer, 14 ... Filter part, 15 ... Weight coefficient selection part.

Claims (2)

A quantization procedure for quantizing the input image data pixel by pixel and converting it to output image data having one of two or more quantization levels;
A weighting factor selection procedure for selecting a weighting factor for error diffusion according to the gradation value of the input image data of the target pixel;
For the pixel of interest, the quantization error generated by the quantization in the quantization procedure is multiplied by the weighting factor selected by the weighting factor selection procedure, and the input image data of the unquantized pixel in the vicinity of the pixel of interest An error diffusion procedure for diffusing , and
In the weighting factor selection procedure,
When the gradation value of the input image data of the target pixel is included in the first range including the minimum value and the maximum value in the gradation value range of the input image data, the plurality of elements And select the first weighting factor
When the gradation value of the input image data of the target pixel is included in a predetermined second range including the minimum value or the maximum value in the gradation value range of the input image data, the first weighting factor Selecting a second weighting factor composed of a plurality of elements including an element different from the element;
When the gradation value of the input image data of the target pixel is included in a range between the first range and the second range, it is based on the first weighting factor and the second weighting factor. An image processing method for selecting a weighting factor having an element obtained by linear interpolation for each element . Quantization means for quantizing input image data for each pixel and converting the output image data into output image data having any one of two or more quantization levels;
A weighting factor selecting means for selecting a weighting factor for error diffusion according to the gradation value of the input image data of the target pixel;
With respect to the input image data of the unquantized pixel in the vicinity of the target pixel by multiplying the quantization error generated by the quantization in the quantization unit with respect to the target pixel by the weighting factor selected by the weighting factor selection unit. An error diffusion means for diffusing , and
The weight coefficient selection means includes
When the gradation value of the input image data of the target pixel is included in the first range including the minimum value and the maximum value in the gradation value range of the input image data, the plurality of elements And select the first weighting factor
When the gradation value of the input image data of the target pixel is included in a predetermined second range including the minimum value or the maximum value in the gradation value range of the input image data, the first weighting factor Selecting a second weighting factor composed of a plurality of elements including an element different from the element;
When the gradation value of the input image data of the target pixel is included in a range between the first range and the second range, it is based on the first weighting factor and the second weighting factor. An image processing apparatus that selects a weighting factor having an element obtained by linear interpolation for each element .

Images