JP2013235403A - Gradation restoration apparatus and program of the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a gradation restoration apparatus capable of restoring an intermediate gradation value accurately.SOLUTION: A gradation restoration apparatus 1 includes: peripheral gradation weight information generation means 10 which generates peripheral gradation weight information by applying an edge preservation type filter for a gradation reduction image; gradation value histogram estimation means 20 which calculates a restoration gradation value of each pixel of an original image by multiplying the restoring magnification by the gradation value of each pixel in a gradation reduction image and estimates a gradation value histogram relative to the number of pixels for the restoration gradation value and the number of pixels for an intermediate gradation value which is between the restoration gradation values; and intermediate gradation restoration image generation means 30 which calculates an alignment sequence by the gradation weight information regarding the pixel of the same gradation value in the gradation reduction image, refers to the gradation value histogram and generates an intermediate gradation restoration image in which the restoration gradation value or the intermediate gradation value is allocated to the pixel of the same gradation value in the calculated alignment sequence.

Description

  The present invention relates to a gradation restoration apparatus for restoring gradations reduced from an original image and a program thereof.

Conventionally, inventions for restoring gradation in image and facsimile vector quantization have been proposed (see Patent Documents 1 and 2).
In the invention described in Patent Document 1, data of the same color component is obtained from a plurality of pixels by extracting data of the same color component from a color image having N gradation values. Then, using the acquired data of the same color component, a color image having M gradation values is restored from a color image having N gradation values (where N <M).

  The invention described in Patent Document 2 uses a so-called “same-direction linearization method” when converting an image into a vector. Specifically, in the invention described in Patent Document 2, gradation points whose vector line directions are in the same event between the characteristic points of gradation change in the main scanning line direction or the sub-scanning line direction are represented by a single line. Represented by a binarized straight line (reference line). Then, the quantized gradation value between each feature point is expressed as a black or white pattern with an area range in which the target gradation value is on the left or right side of the reference line with the reference line as the center, Perform patterning.

Japanese Patent Laid-Open No. 4-45660 (Japanese Patent No. 3089355) JP-A-9-298747 (Japanese Patent No. 3448601)

However, since the inventions described in Patent Documents 1 and 2 do not consider the correlation of gradations in the spatial direction, there is a problem that intermediate gradation values cannot be accurately restored.
Therefore, an object of the present invention is to solve the above-described problem and provide a gradation restoration device and a program thereof that can accurately restore intermediate gradation values.

  In view of the above-described problem, the gradation restoration apparatus according to the first invention of the present application is a gradation restoration apparatus that restores gradations reduced from an original image, and includes gradation weight information generation means, histogram calculation means, And an intermediate gradation restored image generating means.

  According to this configuration, the gradation restoration device receives the gradation reduced image obtained by reducing the gradation of the original image by the gradation weight information generation unit, and applies the edge preserving filter to the input gradation reduced image. Thus, gradation weight information in which the gradation value of each pixel of the gradation reduced image is weighted is generated.

  Here, in a general image, in an edge region such as the boundary between the foreground and the background, the gradation value of each pixel is greatly different on both sides of the edge, and the correlation of gradation is low in the spatial direction. On the other hand, in a general image, in a gradation area such as the inside of a subject, the gradation values of each pixel are smoothly continuous, and the gradation correlation is high in the spatial direction. Therefore, the tone weight information generation means weights the tone values in consideration of spatial continuity of the tone using an edge preserving filter so that the edge region is steep and the gradation region is smooth.

  In addition, the gradation restoration device calculates a gradation-reduced image histogram indicating the number of pixels for each gradation value from the gradation-reduced image by the histogram calculation means, and adds the original gradation value to the calculated gradation-reduced image histogram. The restoration gradation value is calculated by multiplying the ratio of the number of gradations between the image and the gradation-reduced image, and the number of pixels for each restoration gradation value and the number of pixels for each intermediate gradation value between the restoration gradation values are calculated. The gradation value histogram shown is calculated.

  Further, the gradation restoration device calculates the arrangement order based on the gradation weight information for pixels having the same gradation value in the gradation reduced image by the intermediate gradation restoration image generation unit, and refers to the gradation value histogram, Based on the number of pixels of the same gradation value and the ratio of the number of pixels of the restored gradation value and the intermediate gradation value corresponding to the same gradation value, the restored gradation is applied to the pixels of the same gradation value in the calculated arrangement order. An intermediate gradation restored image to which values or intermediate gradation values are assigned is generated. That is, the halftone restored image generation means assigns a restored tone value or an intermediate tone value to each pixel in consideration of spatial continuity of the tone.

Further, in the gradation restoration apparatus according to the second invention of the present application, the gradation weight information generation means uses the gradation value I, the spatial distance D ₁ , the luminance distance D ₂ , the horizontal position i of the pixel, the pixel as the edge preserving filter. The tone weight information O (i, j) is calculated using the Gaussian filter of the formula (1) expressed by the vertical position j, the Gaussian filter region size S, and the Gaussian variance σ. To do.

According to such a configuration, since the gradation restoration apparatus includes the terms of the spatial distance D ₁ and the luminance distance D _{2 in} the expression (1), the spatial continuity of gradation is included in the gradation weight information O (i, j). Can be reflected more accurately.

Further, the gradation restoration apparatus according to the third invention of the present application is characterized in that the histogram calculation means calculates a gradation value histogram by a least square method.
According to such a configuration, the gradation restoration device calculates a gradation value histogram using the property that the gradation values of each pixel are smoothly continuous in a general image.

In the gradation restoration apparatus according to the fourth invention of the present application, the histogram calculation means calculates the gradation value histogram so that the number of pixels of the intermediate gradation value is the same as the number of pixels of the restoration gradation value. Features.
According to such a configuration, the gradation restoration device can calculate the gradation value histogram by a simple calculation process.

  The gradation restoration apparatus according to the first invention of the present application can also be realized by a gradation restoration program that causes a general computer having hardware resources such as a CPU, a memory, and a hard disk to operate cooperatively as the above-described means. . This gradation restoration program may be distributed via a communication line, or may be distributed by writing in a recording medium such as a CD-ROM or a flash memory.

According to the present invention, the following excellent effects can be obtained.
According to the first invention of the present application, gradation weight information in consideration of the spatial continuity of gradation is generated, and a restored gradation value or an intermediate gradation value is assigned to each pixel in the arrangement order of the gradation weight information. As a result of the assignment, the intermediate gradation value can be accurately restored.

According to the second aspect of the present invention, since the spatial continuity of the gradation is more accurately reflected in the gradation weight information, the accuracy of the intermediate gradation value can be further improved.
According to the third aspect of the present invention, since the gradation value histogram in which the number of pixels of the intermediate gradation value is continuously continuous is calculated, the accuracy of the intermediate gradation value can be further improved.
According to the fourth aspect of the present invention, since the gradation value histogram is calculated by a simple calculation process, the calculation process can be speeded up.

It is a block diagram which shows the structure of the gradation restoration apparatus which concerns on 1st Embodiment of this invention. It is a figure explaining calculation of the gradation reduction image histogram by the gradation value histogram estimation means of FIG. It is a figure explaining estimation of the gradation value histogram by the gradation value histogram estimation means of FIG. FIG. 6 is a diagram for explaining a restored gradation value and an intermediate gradation value in the first embodiment of the present invention. 2A and 2B are diagrams for explaining restoration of an intermediate gradation value by the intermediate gradation value restoration unit in FIG. 1, (a) showing a gradation-reduced image, (b) showing peripheral gradation weight information, and (c). An intermediate gradation restoration image is shown. It is a flowchart which shows operation | movement of the gradation decompression | restoration apparatus of FIG. It is a block diagram which shows the structure of the gradation restoration apparatus which concerns on 2nd Embodiment of this invention. It is a figure explaining the production | generation of the gradation value histogram by the gradation value histogram production | generation means of FIG. FIG. 6 is a diagram for explaining restoration of intermediate gradation values in a modification of the present invention, where (a) shows a gradation-reduced image, (b) shows peripheral gradation weight information, and (c) shows intermediate gradation restoration. An image is shown.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings as appropriate. In each embodiment, means having the same function are denoted by the same reference numerals and description thereof is omitted.

(First embodiment)
[Configuration of gradation restoration device]
With reference to FIG. 1, the configuration of the gradation restoration apparatus 1 according to the first embodiment of the present invention will be described.
The gradation restoration device 1 restores gradations reduced from an original image, and includes peripheral gradation weight information generation means (gradation weight information generation means) 10 and gradation value histogram estimation means (histogram calculation means). ) 20 and halftone restored image generating means 30.

Here, the original image is a general image obtained by photographing a person, a landscape, and the like, for example. In addition, a CG image created by CG (computer graphics) is excluded from this original image because the spatial continuity of gradation is poor.
The gradation-reduced image is an image in which the gradation of the original image is reduced using gradation reduction processing such as a truncation method, a rounding off method, and a Lloyd-Max method.
This Lloyd-Max method is described, for example, in the reference ““ A genetic Lloyd-max image quantization algorithm ”, P. Scheumders, Pattern Recognition Letters, Vol. 17, issue 4, p. 547-556. 1996”.

  In this embodiment, the gradation number of the original image is 10 bits (gradation value = 0 to 1023, gradation number = 1024), and the gradation number of the gradation reduced image is 8 bits (gradation value = 0 to 255). , Number of gradations = 256). Then, the gradation restoration device 1 restores the gradation reduced image to the same number of gradations as the original image.

<Peripheral gradation weight information generation means>
The peripheral gradation weight information generation means 10 applies peripheral edge weight information (tone weight information) in which the gradation value of each pixel of the gradation reduced image is weighted by applying an edge preserving filter to the gradation reduced image. ).

First, input / output of the peripheral gradation weight information generation unit 10 will be described.
The peripheral gradation weight information generation unit 10 receives a gradation-reduced image and filter information to be described later from the outside.
The peripheral gradation weight information generation unit 10 outputs the generated peripheral gradation weight information to the intermediate gradation restored image generation unit 30 (peripheral gradation weight information detection unit 33).

Next, the processing of the peripheral gradation weight information generation unit 10 will be described.
In the present embodiment, the peripheral tone weight information generating means 10 uses the bilateral Gaussian filter defined by the following formula (1) as the edge preserving filter, and uses the peripheral tone weight information O ( i, j) is generated. That is, the peripheral gradation weight information generation means 10, the convolution processing using the filter information, according to the spatial distance D ₁ and the luminance distance D ₂ to the gradation value I of each pixel of the gradation reduced image (i, j) Perform weighting. At this time, the peripheral gradation weight information generation unit 10 changes the pixel position (i, j) so as to perform raster scanning, and the peripheral gradation weight information O (i, j) is obtained from all the pixels of the gradation reduced image. ) Is generated.

In this equation (1), the gradation value (luminance value) I, the spatial distance D ₁ , the luminance distance D ₂ , the horizontal position i of the pixel, the vertical position j of the pixel, the size S of the pixel area to which the Gaussian filter is applied, and The Gaussian dispersion value σ.
For example, when the Gaussian filter region size S is 3, m and n take values from -3 to 3.
The filter information includes the Gaussian filter region size S and Gaussian variance values σ ₁ and σ _2, which are set manually. For example, the Gaussian dispersion values σ ₁ and σ ₂ can be set to any value in increments of 1.0 from 0.1 to 10.1, for example.
Hereinafter, the peripheral tone weight information is a value obtained by weighting the tone value of each pixel, and thus may be referred to as a weighted tone value.

<Tone value histogram estimation means>
The gradation value histogram estimation means 20 calculates a gradation reduced image histogram indicating the number of pixels for each gradation value from the gradation reduced image, and multiplies the restoration magnification described later by the gradation value of each pixel of the gradation reduced image. Thus, the restored gradation value is calculated, and the gradation value histogram of the number of pixels of the restored gradation value and the number of pixels of the intermediate gradation value between the restored gradation values is estimated (calculated).

First, input / output of the gradation value histogram estimation means 20 will be described.
This gradation value histogram estimation means 20 receives a gradation reduced image and gradation number information. This gradation number information is information indicating the number of gradations of the original image (for example, 1024) and the number of gradations of the gradation-reduced image (for example, 256), and is manually set.
Further, the gradation value histogram estimation means 20 outputs the estimated gradation value histogram to the intermediate gradation restoration image generation means 30 (intermediate gradation value restoration means 35).

Next, the processing of the gradation value histogram estimation means 20 will be described with reference to FIGS. 2 and 3 (see FIG. 1 as appropriate).
In FIG. 2, the number of pixels of the gradation values ₀ , ₁ , ₂ , and ₃ is indicated by symbols α ₀ , α ₁ , α ₂ , and α ₃ and only a part of the gradation values is illustrated (FIG. 3). , FIG. 4 and FIG. 8 are the same).

The gradation value histogram estimation means 20 refers to the gradation number information, and calculates the ratio of the gradation numbers of the original image and the gradation-reduced image (for example, 1024/256 = 4) as the restoration magnification.
Further, as shown in FIG. 2, the gradation value histogram estimation means 20 calculates a gradation reduced image histogram 90 indicating the number of pixels for each gradation value in the gradation reduced image. In the present embodiment, the number of gradations of the gradation reduced image is 256. Therefore, the gradation value histogram estimation means 20 calculates a gradation reduced image histogram 90 indicating the number of pixels of gradation values = 0, 1, 2, 3,.

Then, as shown in FIG. 3, the gradation value histogram estimation unit 20 calculates the restored gradation value of the gradation value histogram 91 by multiplying the gradation value of the calculated gradation reduced image histogram 90 by the restoration magnification. . In the present embodiment, the restoration magnification is 4, and the gradation values of the gradation-reduced image histogram 90 are 0, 1, 2, 3,. Therefore, the gradation value histogram estimation means 20 calculates restored gradation values = 0, 4, 8, 12,..., 1020 of the gradation value histogram 91. At this time, the number of pixels (vertical axis) of the gradation value histogram 91 becomes the same value as the gradation reduced image histogram 90 of FIG. In other words, code alpha ₀ in FIG. 2, alpha _1, alpha _2, those 4x gradation value (horizontal axis) in alpha _3, code alpha ₀ in FIG. 3, the _{α 1, α 2, α 3} .

Further, the gradation value histogram estimation means 20 estimates (calculates) the number of pixels of the intermediate gradation value by the least square method using the property that the gradation value of each pixel in a general image is smoothly continuous. . In FIG. 3, the intermediate gradation values are 1 to 3 between the restored gradation values 0 and 4, 5 to 7 between the restored gradation values 4 and 8, and the restored gradation values 8 and 12. It will be 9-11 in between. First, the gradation value histogram estimation means 20 estimates the number of pixels having intermediate gradation values 1 to 3 between the restored gradation values 0 and 4 using the least square method. Specifically, the gradation value histogram estimation means 20 draws the least square line β ₁ so that the gradation value histogram 91 connects the number of restored gradation values = 0,4. That is, the least square line β ₁ is obtained by estimating the number of pixels of intermediate gradation values = 1 to 3 between the restored gradation values = 0, 4. Accordingly, tone value histogram estimator 20 estimates the value indicated by the least-squares line beta ₁ as the number of pixels the intermediate tone value = 1-3. Thereafter, the gradation value histogram estimation means 20 subtracts the least square lines β ₂ and β ₃ in the same manner as the least square line β ₁ to obtain the number of pixels of intermediate gradation values = 5 to 7, 9 to 11, and The tone value histogram 91 is estimated.
Needless to say, the gradation value histogram estimation means 20 estimates the number of pixels for all gradation values.

<Intermediate tone restoration image generation means>
Returning to FIG. 1, the description of the configuration of the gradation restoration device 1 will be continued.
The intermediate gradation restored image generation means 30 calculates the arrangement order based on the gradation weight information for pixels having the same gradation value in the gradation reduced image. Then, the halftone restored image generation means 30 refers to the tone value histogram to determine the number of pixels having the same tone value, the number of restored tone values corresponding to the same tone value, and the number of pixels of the halftone value. Based on the ratio, an intermediate gradation restored image is generated by assigning a restored gradation value or an intermediate gradation value to pixels having the same gradation value in the calculated arrangement order. Therefore, the intermediate gradation restoration image generation unit 30 includes a gradation value position information generation unit 31, a peripheral gradation weight information detection unit 33, and an intermediate gradation value restoration unit 35.

First, input / output of the halftone restored image generating means 30 will be described.
The intermediate gradation restored image generation means 30 receives a gradation reduced image, peripheral gradation weight information, gradation number information, and a gradation value histogram.
Further, the halftone restored image generation means 30 outputs the generated halftone restored image to the outside.

Next, with reference to FIGS. 4 and 5, processing of each unit included in the halftone restored image generation unit 30 will be described.
The intermediate gradation restoration image generating means 30 refers to the input gradation number information 93 and sequentially changes from the minimum gradation value (for example, 0) to the maximum gradation value (for example, 255) of the gradation reduced image 92. Designated as the designated gradation value (same gradation value), each means of the intermediate gradation restoration image generation means 30 performs processing on the pixel of the designated gradation value.

  Hereinafter, in order to simplify the description, it is assumed that the designated gradation value = 32 and the gradation-reduced image 92 includes 100 pixels with the designated gradation value = 32. Therefore, as shown in FIG. 4, the restored gradation value is 128, which is the designated gradation value = 32 multiplied by the restoration magnification = 4. Further, the intermediate gradation values are 127, 129, and 130 that are moved back and forth by the gradation value of the restoration magnification = 4 with the restored gradation value = 128 as a reference. In other words, the specified gradation value = 32 in the gradation reduced image 92 corresponds to the gradation values 127 to 130 in the intermediate gradation restored image 94.

  In the gradation value histogram 91 of FIG. 4, the number of pixels with gradation value = 127 is 19, the number of pixels with gradation value = 128 is 29, and the number of pixels with gradation value = 129 is 27. The number of pixels with gradation value = 130 is 25 (the total number of pixels is 100 with gradation values 127 to 130).

In FIG. 5, paying attention to the vicinity of the edge Eg, the gradation value of each pixel is represented by numerical values such as 5, 8, and 32, and the pixel of the designated gradation value = 32 is illustrated by reference symbols a, b, c, and d ( Number with circles).
Further, description will be made by paying attention to four pixels a, b, c, and d out of 100 pixels having designated gradation value = 32.

  As shown in FIG. 5A, the gradation value position information generating unit 31 detects pixels a, b, c, and d having a designated gradation value = 32 from the gradation reduced image 92. The gradation value position information generation unit 31 generates position information (pixel coordinates) of the detected pixels a, b, c, and d, and outputs the position information to the peripheral gradation weight information detection unit 33.

  As shown in FIG. 5B, the peripheral gradation weight information detection unit 33 uses the peripheral gradation weight information 93 for all the input pixels to determine the pixels a, b, c, and d specified by the position information. Weighted gradation values = 33, 37, 34, 28 are extracted. Then, the peripheral gradation weight information detection unit 33 outputs the position information and weighted gradation values (peripheral gradation weight information) of the pixels a, b, c, and d to the intermediate gradation value restoration unit 35.

  As shown in FIG. 5B, the intermediate gradation value restoring unit 35 arranges the pixels a, b, c, and d having the designated gradation value = 32 in the gradation reduced image 92 according to the weighted gradation values. Is calculated. That is, the intermediate gradation value restoring unit 35 rearranges 100 pixels having the designated gradation value = 32 in ascending order of the weighted gradation values, and calculates the arrangement order of the pixels. In FIG. 5B, the arrangement order of the pixels a, b, c, d is illustrated by superscript values, the pixel a is the 42nd, the pixel b is the 87th, the pixel c is the 63rd, Pixel d is 31st.

  Further, as shown in FIG. 5C, the intermediate gradation value restoring means 35 refers to the gradation value histogram 91 of FIG. 4 and designates the specified gradation value = 32 in the order of weighted gradation values. The restored gradation value = 128 or the intermediate gradation value = 127, 129, 130 is assigned to the pixels a, b, c, d. As described above, the total number of pixels having the designated gradation value = 32 is 100. In the gradation value histogram 91 of FIG. 4, the ratio of the number of pixels with gradation values of 127, 128, 129, and 130 is 19: 29: 27: 25. For this reason, the intermediate gradation value restoring means 35 has 19 specified gradation values = 32 and 29 in the ascending order of weighted gradation values for 100 designated gradation values = 32 pixels. The restored gradation value = 128, the intermediate gradation value = 129 for the 27 pixels, and the intermediate gradation value = 130 for the 25 pixels, respectively. Accordingly, gradation values = 128, 130, 129, and 128 are assigned to the pixels a, b, c, and d, respectively.

  After that, each unit of the halftone restored image generation unit 30 performs the process of FIG. 5 on all the pixels of the specified tone value, and assigns tone values = 0 to 1023 to all the pixels. A restored image 94 is generated.

[Operation of gradation restoration device]
The operation of the gradation restoration apparatus 1 will be described with reference to FIG. 6 (see FIG. 1 as appropriate).
The gradation restoration device 1 generates peripheral gradation weight information by applying an edge preserving filter to the gradation-reduced image by the peripheral gradation weight information generation means 10 (step S1).

  The gradation restoration apparatus 1 calculates a gradation reduced image histogram by the gradation value histogram estimation means 20. Further, the gradation restoration device 1 calculates a restored gradation value by using the gradation value histogram estimation means 20, estimates an intermediate gradation value, and estimates a gradation value histogram (step S2).

In the gradation restoration device 1, an intermediate gradation restored image is generated by the intermediate gradation restored image generation means 30.
The gradation value position information generation unit 31 detects pixels having a designated gradation value from the gradation reduced image, and generates position information of the detected pixels (step S3).
The peripheral gradation weight information detection unit 33 extracts the weighted gradation value of the pixel specified by the position information from the peripheral gradation weight information of all the pixels (step S4).
The intermediate gradation value restoration unit 35 calculates the arrangement order of the weighted gradation values for the pixels having the designated gradation value in the gradation reduced image, and refers to the gradation value histogram to determine the weighted gradation value. The restored gradation value or the intermediate gradation value is assigned to the pixels having the designated gradation value in the arrangement order (step S5).

  As described above, the gradation restoration apparatus 1 according to the first embodiment of the present invention generates gradation weight information in consideration of the spatial continuity of gradations, and in the order in which the weighted gradation values are arranged. Since the restored gradation value or the intermediate gradation value is assigned to each pixel, the intermediate gradation value can be accurately restored.

  Furthermore, since the tone restoration apparatus 1 more accurately reflects the spatial continuity of the tone in the tone weight information by the Gaussian filter of Expression (1), the accuracy of the intermediate tone value can be further improved. it can.

  Furthermore, the gradation restoration apparatus 1 estimates the number of pixels of the intermediate gradation value by the method of least squares using the characteristic that the gradation value of each pixel is smoothly continuous in a general image. For this reason, the gradation restoring apparatus 1 can smoothly improve the accuracy of the intermediate gradation value by smoothly smoothing the number of pixels of the intermediate gradation value in the gradation value histogram.

  Here, since the pixel density of the ultra-high-definition image is fine, there is a high possibility that a gradation value at a portion having a high spatial frequency in an image having a normal definition becomes an intermediate gradation value. For this reason, the gradation restoration device 1 is particularly effective for an ultra-high definition image.

(Second Embodiment)
With reference to FIG. 7, a difference from the first embodiment will be described regarding the gradation restoration device 1 </ b> A according to the second embodiment of the present invention.
As shown in FIG. 7, the gradation restoration apparatus 1A includes a gradation value histogram generation means (histogram calculation means) 20A instead of the gradation value histogram estimation means 20 of FIG.
This gradation value histogram generation means 20A is different from the gradation value histogram estimation means 20 of FIG. 1 in that the intermediate gradation value is copied (calculated) as the same value as the restored gradation value.

<Tone value histogram generating means>
With reference to FIG. 8, the processing of the gradation value histogram generation means 20A will be described (see FIG. 7 as appropriate).
The gradation value histogram generation unit 20A calculates a gradation-reduced image histogram 90 (FIG. 2) and calculates a restored gradation value, similarly to the gradation value histogram estimation unit 20 of FIG. In other words, code alpha ₀ in FIG. 2, alpha _1, alpha _2, is obtained by four times the gradation value in the alpha _3, code alpha ₀ in FIG. 8, alpha _1, alpha _2, the alpha _3.

  As shown in FIG. 8, the gradation value histogram generating unit 20A copies (calculates) the intermediate gradation value as the same value as the restored gradation value. For example, attention is paid to intermediate gradation values = 3, 5, and 6 that are restored by a gradation value of restoration magnification = 4 with the restoration gradation value 4 as a reference. In this case, the gradation value histogram generation unit 20A copies the number of pixels of the restored gradation value 4 as the number of pixels of intermediate gradation values = 3, 5, and 6. That is, in the gradation value histogram 91A, the number of pixels of gradation value = 3 to 6 is the same value.

Similarly, the gradation value histogram generation means 20A copies the number of pixels with the restored gradation value = 8 as the number of pixels with the intermediate gradation value = 7, 9, 10 and sets the number of pixels with the restored gradation value = 12. Copying is performed as the number of pixels of the intermediate gradation value = 11, 13, and 14. In this way, the gradation value histogram generation unit 20A generates a gradation value histogram 91A indicating the number of pixels of all gradation values.
Note that the gradation value histogram generating unit 20A copies the number of pixels with the restored gradation value 0 as the number of pixels with the intermediate gradation values 1 and 2 because there is no negative gradation value in the gradation value histogram 91A.

  As described above, the gradation restoration device 1A according to the second embodiment of the present invention can accurately restore the intermediate gradation values as in the first embodiment. Further, the gradation restoration device 1A can increase the speed of the arithmetic processing because the gradation value histogram generating means 20A generates the gradation value histogram by a simple arithmetic processing.

  As mentioned above, although embodiment of this invention was described, this invention is not limited to this, It can implement in the range which does not change the meaning. The modification of embodiment is shown below.

(Modification 1)
In each embodiment, it has been described that the Gaussian filter of Expression (1) is used, but the edge preserving filter of the present invention is not limited to this.
That is, the peripheral tone weight information generating means 10 can also use a monolateral Gaussian filter defined by the following equation (2) as the edge preserving filter. The equation (2) is obtained by deleting the section on luminance distance D ₂ from equation (1).

<Comparison of halftone restored images>
Even when the Gaussian filter of Expression (2) is used, each unit of the halftone restored image generation unit 30 generates the halftone restored image 94A by the same processing as in the first embodiment. However, the generated halftone restored image 94A has different tone values from the halftone restored image 94 because the Gaussian filters are different as in the equations (1) and (2). Therefore, with reference to FIG. 9, the difference between the halftone restored images generated by the Gaussian filters of Expressions (1) and (2) will be described (see FIGS. 1 and 5 as appropriate).

  The gradation value position information generation unit 31 detects pixels a, b, c, and d having a designated gradation value = 32 from the gradation reduced image 92 as shown in FIG. This processing result is not different from that shown in FIG.

  As shown in FIG. 9B, the peripheral gradation weight information detecting unit 33 calculates the weighted gradation values 33, 34, and b of the pixels a, b, c, and d from the peripheral gradation weight information 93A of all the pixels. 33 and 19 are extracted. That is, in FIGS. 5B and 9B, since the Gaussian filters are different as in Expression (1) and Expression (2), the weighted gradation values of the pixels a, b, c, and d are different. Yes.

  As shown in FIG. 9B, the intermediate tone value restoring means 35 calculates the order of weighted tone values for the pixels a, b, c, and d, and uses the order of weighted tone values. The restored gradation value = 128 or the intermediate gradation value = 127, 129, 130 is assigned to the pixels a, b, c, d. In FIG. 9B, the pixel a is the 36th, the pixel b is the 65th, the pixel c is the 37th, and the pixel d is the 4th. Accordingly, the gradation values 128, 129, 129, and 127 are assigned to the pixels a, b, c, and d, respectively.

  Here, in the intermediate gradation restoration image 94 of FIG. 5C, the gradation values of the pixels b, c, and d = 130, 129, and 128. On the other hand, in the intermediate gradation restored image 94A of FIG. 9C, the gradation values of the pixels b, c, and d = 129, 128, 127. That is, the gradation values of the pixels b, c, and d are higher in the intermediate gradation restored image 94 than in the intermediate gradation restored image 94A, and the difference between the gradation values of the pixels located on the right side of the edge Eg is larger. . From the above, it can be seen that the edge Eg is sufficiently clear in the intermediate gradation restored image 94A, and the edge Eg is extremely clear in the intermediate gradation restored image 94.

(Other variations)
In each embodiment, the gradation is linearly restored, but the present invention is not limited to this.
That is, the gradation restoration apparatuses 1 and 1A can perform nonlinear quantization such as a Lloyd-max method. In this case, the gradation restoration apparatuses 1 and 1A store in advance a code book indicating the correspondence between the gradation value of the gradation reduced image and the gradation value of the intermediate gradation restored image, and refer to this code book. Thus, the range of the intermediate gradation value to be restored may be determined.

  Each embodiment has been described as including either the gradation value histogram estimation means 20 or the gradation value histogram generation means 20A, but the present invention is not limited to this. That is, the gradation restoration apparatus according to the present invention includes both the gradation value histogram estimation means 20 and the gradation value histogram generation means 20A, and it is possible to manually set which means calculates the gradation value histogram. Good.

1,1A gradation restoration device 10 peripheral gradation weight information generation means (gradation weight information generation means)
20 gradation value histogram estimation means (histogram calculation means)
20A gradation value histogram generating means (histogram calculating means)
30 Intermediate tone restoration image generation means 31 Tone value position information generation means 33 Peripheral gradation weight information detection means 35 Intermediate gradation value restoration means

Claims (5)

A gradation restoration device for restoring gradations reduced from an original image,
A gradation-reduced image obtained by reducing the gradation of the original image is input, and a gradation value of each pixel of the gradation-reduced image is weighted by applying an edge preserving filter to the input gradation-reduced image. Gradation weight information generating means for generating the specified gradation weight information;
A gradation-reduced image histogram indicating the number of pixels for each gradation value is calculated from the gradation-reduced image, and the gradation values of the calculated gradation-reduced image histogram are added to the levels of the original image and the gradation-reduced image. The restoration gradation value is calculated by multiplying the ratio of the logarithm, and the gradation value histogram indicating the number of pixels for each restoration gradation value and the number of pixels for each intermediate gradation value between the restoration gradation values is calculated. Histogram calculating means for
For the pixels with the same gradation value in the gradation-reduced image, the arrangement order based on the gradation weight information is calculated, and the number of pixels with the same gradation value and the same gradation are calculated with reference to the gradation value histogram. Based on the restored gradation value corresponding to the value and the ratio of the number of pixels of the intermediate gradation value, the restored gradation value or the intermediate gradation value is assigned to the pixels having the same gradation value in the calculated arrangement order. Halftone restored image generating means for generating a halftone restored image;
A gradation restoration apparatus comprising: The gradation weight information generating means is an edge-preserving filter that has a gradation value I, a spatial distance D ₁ , a luminance distance D ₂ , a horizontal position i of the pixel, a vertical position j of the pixel, and a Gaussian filter area size. _{2. The} gradation weight information O (i, j) is calculated using a Gaussian filter of Expression (1) expressed by S and Gaussian dispersion values σ ₁ and σ _2. The gradation restoration apparatus described.

  The gradation restoration apparatus according to claim 1, wherein the histogram calculation unit calculates the gradation value histogram by a least square method.   The histogram calculation means calculates the gradation value histogram so that the number of pixels of the intermediate gradation value is the same as the number of pixels of the restored gradation value. 2. The gradation restoration apparatus described in 1. In order to restore the reduced gradation from the original image,
A gradation-reduced image obtained by reducing the gradation of the original image is input, and a gradation value of each pixel of the gradation-reduced image is weighted by applying an edge preserving filter to the input gradation-reduced image. Gradation weight information generating means for generating received gradation weight information,
A gradation-reduced image histogram indicating the number of pixels for each gradation value is calculated from the gradation-reduced image, and the gradation values of the calculated gradation-reduced image histogram are added to the levels of the original image and the gradation-reduced image. The restoration gradation value is calculated by multiplying the ratio of the logarithm, and the gradation value histogram indicating the number of pixels for each restoration gradation value and the number of pixels for each intermediate gradation value between the restoration gradation values is calculated. Histogram calculation means to
For the pixels with the same gradation value in the gradation-reduced image, the arrangement order based on the gradation weight information is calculated, and the number of pixels with the same gradation value and the same gradation are calculated with reference to the gradation value histogram. Based on the restored gradation value corresponding to the value and the ratio of the number of pixels of the intermediate gradation value, the restored gradation value or the intermediate gradation value is assigned to the pixels having the same gradation value in the calculated arrangement order. Intermediate gradation restored image generating means for generating a restored intermediate gradation image,
Gradation restoration program to function as

Images