JP6164977B2 - Image processing apparatus, image processing method, and image processing program - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program that output a flatness index value that is an index of flatness in a predetermined region of an input image.

Since the subsequent image processing can be simplified while retaining the structural attribute of the image, for example, Japanese Unexamined Patent Application Publication No. 2008-293425 (Patent Document 1) has been proposed as a technique relating to flatness determination.
That is, in Patent Document 1, the sum of absolute value differences between a target pixel and a plurality of peripheral pixels centered on the target pixel is calculated for each of a plurality of directions, and the maximum value among them is adopted to calculate the target pixel. A technique for determining the flatness of the image is disclosed.

JP 2008-293425 A

However, in the technique disclosed in Patent Document 1 described above, a low-contrast edge or texture in a high-sensitivity photographed image is very similar to the amplitude of noise, and a pixel of interest and a plurality of peripheral pixels in an area where the edge or texture exists. There is no clear difference between the sum of absolute differences between and the sum of absolute differences between the pixel of interest and a plurality of peripheral pixels in a flat portion including noise (a region where no clear edge or texture exists). For this reason, in the high-sensitivity photographed image, the accuracy of flatness determination for determining whether or not a certain region is flat is reduced.
On the other hand, a method of determining whether or not the image is flat while averaging and suppressing the influence of noise by setting and processing a region of interest including pixels existing around each pixel can be considered. However, in this case, setting the processing target area may cause a high-contrast portion to enter the area, thereby reducing the accuracy of discrimination between the low-contrast area and the flat area.

  The present invention has been made in view of the above-described circumstances, and an object thereof is to improve the accuracy of discrimination between a low-contrast region and a flat region without being affected by amplitude due to noise.

In order to achieve the above object, the present invention provides the following means.
According to an aspect of the present invention, a processing region setting unit that determines a processing region that is a processing target region from an input image, and a region centered on the target pixel in the processing region is extracted as a target block. Based on a difference between pixel values of a pixel included in the target block and a pixel included in the reference block; The similarity calculation means for calculating the similarity between the target pixel and the reference pixel, and the reference pixel according to the similarity, the higher the similarity, the lower the similarity. Based on the weighting means for calculating the weight value, the pixel value of each pixel in the processing region, and the weight value for the reference pixel, the flatness of the processing region To provide an image processing apparatus comprising: a flatness index value calculating means for calculating a flatness index value as a target, a.

According to this aspect, the processing area setting means sets the area to be processed in the input image as the processing area, and in this processing area, the block extraction means extracts the area centered on the target pixel as the target block. At the same time, an area having the same size as the target block is extracted as a reference block with a reference pixel different from the target pixel as a center. Then, the similarity calculation means calculates the similarity between the target pixel and the reference pixel for each pixel in the processing region based on the difference in pixel values between the pixel included in the target block and the pixel included in the reference block. The weight value of each pixel is calculated according to the similarity by the weighting means. As a result, when the target pixel has low contrast, the contribution ratio of the high-contrast portion with low similarity to the target pixel can be reduced to calculate a flatness index value that is an index of the flatness of the processing region. it can.
In this way, the flatness index value is calculated without mixing high-contrast edges and textures in the processing target area, or even if they are mixed, so that the flatness index value is calculated. Can be improved.

In the aspect described above, the flatness index calculating unit includes a feature amount calculating unit that calculates a feature amount corresponding to an amplitude of a high frequency component of each pixel in the processing region, and weights the feature amount according to the weight value. The flatness index value is preferably calculated.
In this way, even when noise having a large amplitude is included in the low contrast portion in the processing region, it is possible to improve the discrimination accuracy between the flat region and the low contrast region.

In the aspect described above, the flatness index value calculating unit calculates a root mean square of pixel values between adjacent pixels in a plurality of directions of the processing region for each direction, and the maximum value among these is calculated as a flatness index It is preferable to use a value.
In this way, even when noise having a large amplitude is included in the low contrast portion in the processing region, it is possible to improve the discrimination accuracy between the flat region and the low contrast region.

In the aspect described above, the feature amount calculation unit calculates a value based on a difference in pixel values between adjacent pixels in each of a plurality of predetermined directions for each pixel in the processing region, It is preferable that the maximum value based on the difference for each direction is the feature amount of each pixel.
In this way, even when noise having a large amplitude is included in the low contrast portion in the processing region, it is possible to improve the discrimination accuracy between the flat region and the low contrast region.

In the above aspect, the weighting unit may set the weight value to 1 when the similarity is smaller than a predetermined threshold, and set the weight value to zero when the similarity is larger than the threshold. preferable.
In this way, since the flatness index value is calculated by excluding the high contrast region from the processing region, it is possible to easily improve the discrimination accuracy between the flat region and the low contrast region.

In the aspect described above, it is preferable that the weighting unit corrects the weight value based on edge information indicating edge information of the input image.
In the above aspect, the weighting unit calculates edge information indicating a degree of edge in each pixel of the input image, and the weight value is decreased when the edge information value of the reference pixel is high. It is preferable to correct the weight value.
By doing so, it is possible to improve the accuracy of discrimination between the low contrast region and the flat region without being affected by noise, shadows, brightness variations, and the like.

In the above aspect, when the input image is an image signal in which a specific color component is missing, color component determination means that determines a color component in each pixel of the processing region and outputs it as color information,
The similarity calculation means calculates, for each color information, a difference between a pixel value of a pixel included in the target block and a pixel included in the reference block, and based on the difference, between the target pixel and the reference pixel It is preferable to calculate the similarity.
By doing in this way, since the information regarding the presence or absence of the structure for each color component can be obtained, the flatness index value can be calculated more accurately.

  Further, according to one aspect of the present invention, a processing region setting step that determines a processing region that is a processing target region from the input image, and in the processing region, a region centered on the target pixel is extracted as a target block, A block extraction step of extracting a reference block having an area of the same size as the target block with a reference pixel different from the target pixel as a center, and a pixel value difference between the pixel included in the target block and the pixel included in the reference block; Based on the similarity calculation step for calculating the similarity between the target pixel and the reference pixel based on the reference pixel, the higher the similarity, the lower the similarity. Based on the weighting step of calculating a lower weight value, the pixel value of each pixel in the processing region, and the weight value for the reference pixel, A flatness index value calculation step of calculating a flatness index value as an index of the flatness of the serial processing region, to provide an image processing method comprising.

  Further, according to one aspect of the present invention, a processing region setting step that determines a processing region that is a processing target region from the input image, and in the processing region, a region centered on the target pixel is extracted as a target block, A block extraction step of extracting a reference block having an area of the same size as the target block with a reference pixel different from the target pixel as a center, and a pixel value difference between the pixel included in the target block and the pixel included in the reference block; Based on the similarity calculation step for calculating the similarity between the target pixel and the reference pixel based on the reference pixel, the higher the similarity, the lower the similarity. A weighting step for calculating a lower weight value, a pixel value of each pixel in the processing region, and the weight value for the reference pixel To provide an image processing program for executing a calculation step flatness index value calculating a flatness flatness index value indicative of the processing area in a computer.

  According to the present invention, it is possible to improve the discrimination accuracy between a flat region and a low contrast region without being affected by the amplitude due to noise.

1 is a block diagram illustrating a schematic configuration of an image processing apparatus according to a first embodiment of the present invention. It is explanatory drawing which shows the extraction example of an attention block and a reference block in the 1st Embodiment of this invention. In the 1st Embodiment of this invention, it is a reference figure which shows the calculation direction at the time of calculating a flatness index value. In the 1st Embodiment of this invention, it is explanatory drawing which shows the example at the time of producing | generating mask information by making a weight value into 0 or 1. FIG. FIG. 5 is a reference diagram illustrating a calculation direction when generating mask information and calculating a flatness index value in the first embodiment of the present invention. It is a block diagram which shows schematic structure of the image processing apparatus which concerns on the 2nd Embodiment of this invention. It is explanatory drawing which shows the extraction example of an attention block and a reference block in the 2nd Embodiment of this invention.

(First embodiment)
The image processing apparatus according to the first embodiment of the present invention will be described below with reference to the drawings.
As shown in FIG. 1, the image processing apparatus includes a processing region setting unit 11 that determines a predetermined region to be processed in an input image, a block extraction unit 12 that extracts a target block and a reference block, and a similarity of a reference pixel to a target pixel. A similarity calculation unit 13 that calculates a weight, a weighting unit 14 that calculates a weighting value according to the similarity, and a flatness index value calculation unit 15 that calculates a flatness index value that is an index of the flatness of the processing region. .

  The processing area setting unit 11 determines a predetermined area to be processed in the input image. For example, as shown in FIG. 2, a 9 × 9 rectangular area centered on the target pixel is set as the processing area. Subsequent processing is performed in this processing region, and a flatness index value serving as an index indicating the flatness of the processing region centered on the target pixel is calculated.

  In this specification, the flatness is calculated for each pixel of the input image and indicates the degree to which the region including the target pixel is flat. When the flatness of the region including the target pixel is high, it can be determined that there is no edge (or few) around the target pixel. On the other hand, when the flatness is low, the degree of flatness of the target pixel is low, and it can be determined that there is a pattern such as a fine texture or an edge. In this embodiment, the flatness index value is calculated such that the higher the flatness, the lower the flatness, and the higher the flatness, the higher the flatness index value.

  In the processing region set by the processing region setting unit 11, the block extraction unit 12 extracts a predetermined region centered on the target pixel as the target block, and is the same as the target block centered on a reference pixel different from the target pixel. The size area is extracted as a reference block.

  In the example of FIG. 2, a rectangular area of 3 pixels × 3 pixels centered on the target pixel is set as the target block TP1, and a rectangular area of 3 pixels × 3 pixels centered on a reference pixel other than the target pixel is a reference block. RP1 to RP8. FIG. 2 shows an example in which eight reference blocks are set around the block of interest. For example, a plurality of reference blocks can be set by shifting the reference pixels one pixel at a time.

  The similarity calculation unit 13 calculates a similarity index value representing the similarity between the target pixel and the reference pixel based on the difference in pixel value between the pixel included in the target block and the pixel included in the reference block. For example, a difference average between a pixel included in the block of interest and a pixel included in the reference block, a difference square average, a difference absolute value average, or the like can be used as the similarity index value indicating the similarity. When using these similarity index values, the higher the similarity index value, the lower the similarity, and the lower the similarity index value, the higher the similarity.

  Here, in the case of the example of FIG. 2 where the similarity is the mean square difference, the calculation of the similarity index value D1 representing the similarity between the target pixel of the target block TP1 and the reference pixel of the reference block RP1 is as follows ).

注目画素とその他の参照ブロックＲＰ２〜８における参照画素の類似度を表す類似度指標値Ｄ２〜Ｄ８も同様に算出する。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１〜Ｄ８は最大値の類似度指標（ＭＡＸ＿Ｄ）で正規化される。

Similarity index values D2 to D8 representing the similarity between the target pixel and the reference pixels in the other reference blocks RP2 to RP8 are calculated in the same manner. Further, the similarity index values D1 to D8 of TP1 and RP1 to RP8 are normalized with the maximum similarity index (MAX_D).

注目画素とその他の参照ブロックＲＰ２〜８における参照画素の類似度指標値Ｄ２〜Ｄ８も同様に算出する。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１〜Ｄ８は最大値の類似度指標値（ＭＡＸ＿Ｄ）で正規化される。 When the difference average is used as the similarity index value, it is calculated by the following equation (2).

Similarity index values D2 to D8 of the target pixel and reference pixels in the other reference blocks RP2 to RP8 are calculated in the same manner. Further, the similarity index values D1 to D8 of TP1 and RP1 to RP8 are normalized by the maximum similarity index value (MAX_D).

注目画素とその他の参照ブロックＲＰ２〜８における参照画素の類似度指標値Ｄ２〜Ｄ８も同様に算出する。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１〜Ｄ８は最大値の類似度指標値（ＭＡＸ＿Ｄ）で正規化される。 When the difference absolute value average is used as the similarity, it is calculated by the following equation (3).

Similarity index values D2 to D8 of the target pixel and reference pixels in the other reference blocks RP2 to RP8 are calculated in the same manner. Further, the similarity index values D1 to D8 of TP1 and RP1 to RP8 are normalized by the maximum similarity index value (MAX_D).

  The weighting unit 14 calculates the weight value of each reference pixel according to the similarity calculated by the similarity calculation unit 13. At this time, a higher weight value is calculated as the similarity of each reference pixel is higher and lower as the similarity is lower. That is, the reference pixel similar to each target pixel is calculated so that the weight value is high, and the reference pixel not similar is calculated so that the weight value is low.

The flatness index value calculation unit 15 sets a feature amount corresponding to the amplitude of the high-frequency component of each pixel in the processing region, and weights and averages the feature amount according to the weight value (W) calculated by the weighting unit. The flatness index value serving as the index of is calculated.
For example, in a plurality of directions of the processing area, a weighted difference mean square of pixel values between adjacent pixels is calculated for each direction with respect to the weight value (W), and the maximum value among these is used as the flatness index value. .

Specifically, for the weight W (x, y) for each pixel P (x, y), as shown in FIG. 3, specific directions (in FIG. 3, four directions of vertical, horizontal, right diagonal, and left diagonal). The square of the difference between the pixels existing in is calculated and the weighted average is obtained. At this time, a portion where a square of difference cannot be calculated between pixels such as a corner of the processing region is excluded from processing.
For example, the difference square average Dh in the horizontal direction is calculated by the following equation (4).
Dh = ΣΣ (P (x, y) −P (x + 1, y)) ² * W (x, y) / (N * M)
... (4)
(X = 0 to N, y = 0 to M: However, in the case of a pixel existing at coordinates where a difference value cannot be calculated, it is skipped)
The other three directions are calculated in the same manner, and the maximum value among them is used as the flatness index value.

Alternatively, a feature amount for each of the vertical, horizontal, right diagonal, and left diagonal directions, for example, the difference square D of adjacent pixel values may be obtained first for each pixel in the processing region, and the maximum value may be weighted averaged. .
That is, the calculation of the difference square D of adjacent pixel values is calculated by the following equation (4) ′.

  In the above example, the root mean square of pixel values between adjacent pixels is calculated, but is not limited to between adjacent pixels. Further, the mean square difference is used as the flatness index value. However, the present invention is not limited to this, and a value based on another difference, for example, a difference average or a difference absolute value average may be used as the flatness index value. Furthermore, not only the difference of the pixel value between pixels but the value based on the ratio of the pixel value between pixels can also be used.

When the difference average is used as the flatness index value, it is calculated as in the following equation (5).
Dh = ΣΣ (P (x, y) −P (x + 1, y)) * W (x, y) / (N * M)
... (5)
(X = 0 to N, y = 0 to M: However, the calculation of the difference value is skipped for pixels existing at coordinates where the difference value cannot be calculated)

Further, when the difference absolute value average is used as the flatness index value, it is calculated as in the following equation (6).
Dh = ΣΣ | P (x, y) −P (x + 1, y) | * W (x, y) / (N * M)
... (6)
(X = 0 to N, y = 0 to M: However, the calculation of the difference value is skipped for pixels existing at coordinates where the difference value cannot be calculated)

  As described above, according to the present embodiment, when the target pixel has low contrast, the flatness that serves as an index of the flatness of the processing region is reduced by reducing the contribution ratio of the high-contrast portion having low similarity to the target pixel. An index value can be calculated. Accordingly, the flatness index value is calculated without mixing high-contrast edges and textures in the processing target area, or even if they are mixed, and the flatness index value is calculated. Can be improved.

It should be noted that the weight value (W) by the weighting unit 14 is calculated as the following equation (7) according to the similarity index value.
W = exp (−D1 / MAX_D) (7)

In addition, a threshold value can be set in advance, and the weight value can be determined by comparing the similarity index value with the threshold value.
In particular, for each pixel, when the similarity index value is lower than the threshold value, the weight value is set to 1. When the similarity is higher than the threshold value, the weight value is set to zero, thereby calculating the flatness index value. It can be simplified. That is, in this case, as shown in FIG. 4, it is possible to generate mask information that excludes pixels having a weight value of zero from being processed.

  In the case where the target block is a flat part or a low contrast part, a reference pixel having a high similarity, that is, a similarity index value lower than a threshold value, is highly likely to be a flat part or a low contrast part. For this reason, a reference pixel having a high similarity is set as a target pixel when calculating the flatness index. On the other hand, a reference pixel having a low similarity, that is, a similarity index value higher than a threshold is highly likely not to be a flat portion or a low contrast portion. For this reason, the reference information with a low similarity is not set as a target pixel when the flatness index is calculated, and mask information is generated as a mask target.

In the example of FIG. 4, pixels shown in black are pixels that are not processed, and pixels shown in white are processed pixels.
Then, by applying this mask information to the processing region, as shown in FIG. 5, the calculation target pixel is limited when calculating the flatness index value, and the flatness index value is obtained only from the target pixel. calculate.

  By doing so, the flatness index value is calculated without mixing high-contrast edges and textures in the processing target area, or even if they are mixed, and the flatness index value is calculated. In addition to improving the discrimination accuracy with respect to the flat region, it is possible to reduce the number of calculation target pixels when calculating the flatness index value, thereby simplifying the calculation.

(Modification)
The weighting unit 14 according to the first embodiment described above may correct the weight value (W) using edge information (E) indicating the degree of edge of the input image. In other words, correction may be performed such that a weight is reduced for a pixel determined to be an edge and a weight is increased for a pixel determined to be a non-edge.

In this case, any known method can be used as the edge information (E).
For example, the corrected weight value (W_C) is obtained by the following equation (8). Note that the edge information is normalized with the maximum edge information signal.
W_C = W * exp (−E) (8)

As another method, as the similarity between the edge information (E) of the pixel of interest and the edge information (E) of the surrounding reference pixels is lower (the difference value (E_dif) of the edge information (E) is higher), the weight value (W ) Is lowered, and the weight value (W) is increased as the similarity is higher (the difference area (E_dif) of the edge information (E) is lower).
Actual calculation is performed by the following equation (9).
W_C = W * exp (−E_dif) (9)

The corrected weight value (W_C) obtained by the above equations (8) and (9) is replaced with the weight values (W) of the above equations (5), (6), and (7).
Further, when the mask information is generated as described above, for example, when a pixel that is a processing target pixel in the mask information is an edge pixel, correction may be performed so that the pixel is not a processing target pixel. it can.

(Second Embodiment)
Hereinafter, a second embodiment of the present invention will be described. The same components as those in the first embodiment described above are denoted by the same reference numerals, and the description thereof is omitted.
As shown in FIG. 6, the present embodiment is a color component determination unit that determines a color component in each pixel in the processing region and outputs it as color information when the input image is an image signal lacking a specific color component. It has.

  The similarity calculation unit 13 calculates, for each color information output from the color component determination unit 18, a difference between the pixel values of the pixel included in the target block and the pixel included in the reference block, and based on these differences, the target pixel And the similarity between the reference pixel and the reference pixel are calculated.

  For example, as illustrated in FIG. 7, the block extraction unit 12 has a 5-pixel × 5-pixel target block TP1 centered on the target pixel and a 5-pixel × 5 centered on a reference pixel different from the target pixel in the processing region. Pixel reference blocks RP1 to RP8 are extracted. In the example of FIG. 7, the color component type of the input image is RGB.

  Based on the color information from the color component determination unit 18, the similarity calculation unit 13 calculates, for each color information, a pixel value difference between a pixel included in the target block and a pixel included in the reference block. Based on the difference, the similarity between the target pixel and the reference pixel is calculated.

  As the similarity index value indicating the similarity, for example, a mean square difference of each color is obtained for each color information. That is, in the example of FIG. 7, the root mean square difference DR of the R component, the root mean square difference DGr of the Gr component, the root mean square DGb of the Gb component, and the root mean square DB of the B component are calculated. Then, the mean square difference is added to obtain the similarity index value of the reference pixel. When using these similarity index values, the higher the similarity index value, the lower the similarity, and the lower the index value index value, the higher the similarity.

  In the case of FIG. 7, the sum of the root mean square of the target block TP1 and the reference block RP1, that is, the similarity index value D1 of the reference pixel that is the center pixel of the reference block RP1 is expressed by the following equations (10) to (14) ) Easily calculated according to the equation.

注目ブロックＴＰ１とその他の参照ブロックＲＰ２〜ＲＰ８との差分二乗平均の和、つまり他の参照画素の類似度指標値も同様に算出される。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１Ｒ〜Ｄ８Ｒ、Ｄ１Ｇｂ〜Ｄ８Ｇｂ、Ｄ１Ｇｒ〜Ｄ８Ｇｒ、Ｄ１Ｂ〜Ｄ８Ｂは夫々の類似度指標値の最大値（ＭＡＸ＿ＤＲ、ＭＡＸ＿ＤＧｂ、ＭＡＸ＿ＤＧｒ、ＭＡＸ＿ＤＢ）で正規化される。

The sum of the root mean square difference between the target block TP1 and the other reference blocks RP2 to RP8, that is, the similarity index value of other reference pixels is calculated in the same manner. Further, the similarity index values D1R to D8R, D1Gb to D8Gb, D1Gr to D8Gr, and D1B to D8B of TP1 and RP1 to 8 are normalized with the maximum values (MAX_DR, MAX_DGb, MAX_DGr, MAX_DB) of the respective similarity index values. The

注目ブロックＴＰ１とその他の参照ブロックＲＰ２〜ＲＰ８との差分平均の和、つまり他の参照画素の類似度指標値も同様に算出される。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１Ｒ〜Ｄ８Ｒ、Ｄ１Ｇｂ〜Ｄ８Ｇｂ、Ｄ１Ｇｒ〜Ｄ８Ｇｒ、Ｄ１Ｂ〜Ｄ８Ｂは夫々の類似度指標値の最大値（ＭＡＸ＿ＤＲ、ＭＡＸ＿ＤＧｂ、ＭＡＸ＿ＤＧｒ、ＭＡＸ＿ＤＢ）で正規化される。 In addition, you may use another parameter | index as a similarity index value, for example, when the difference average is used for a similarity index value, it calculates with the following (15) Formula-(19) Formula.

The sum of the average difference between the target block TP1 and the other reference blocks RP2 to RP8, that is, the similarity index value of other reference pixels is calculated in the same manner. Further, the similarity index values D1R to D8R, D1Gb to D8Gb, D1Gr to D8Gr, and D1B to D8B of TP1 and RP1 to 8 are normalized with the maximum values (MAX_DR, MAX_DGb, MAX_DGr, MAX_DB) of the respective similarity index values. The

注目ブロックＴＰ１とその他の参照ブロックＲＰ２〜ＲＰ８との差分絶対値平均の和、つまり他の参照画素の類似度指標値も同様に算出される。さらに、ＴＰ１とＲＰ１〜８の類似度指標値Ｄ１Ｒ〜Ｄ８Ｒ、Ｄ１Ｇｂ〜Ｄ８Ｇｂ、Ｄ１Ｇｒ〜Ｄ８Ｇｒ、Ｄ１Ｂ〜Ｄ８Ｂは夫々の類似度指標値の最大値（ＭＡＸ＿ＤＲ、ＭＡＸ＿ＤＧｂ、ＭＡＸ＿ＤＧｒ、ＭＡＸ＿ＤＢ）で正規化される。 Further, when the difference absolute value average is used as the similarity index value, it is calculated according to the following equations (20) to (24).

The sum of absolute difference averages between the target block TP1 and the other reference blocks RP2 to RP8, that is, the similarity index value of other reference pixels is calculated in the same manner. Further, the similarity index values D1R to D8R, D1Gb to D8Gb, D1Gr to D8Gr, and D1B to D8B of TP1 and RP1 to 8 are normalized with the maximum values (MAX_DR, MAX_DGb, MAX_DGr, MAX_DB) of the respective similarity index values. The

  As described above, according to the present embodiment, when the target pixel has low contrast, the flatness that serves as an index of the flatness of the processing region is reduced by reducing the contribution ratio of the high-contrast portion having low similarity to the target pixel. An index value can be calculated. Accordingly, the flatness index value is calculated without mixing high-contrast edges and textures in the processing target area, or even if they are mixed, and the flatness index value is calculated. Can be improved. In particular, information regarding the presence or absence of a structure for each color component can be obtained, and the flatness index value can be calculated more accurately.

  In the above-described embodiments, the image processing apparatus has been described on the premise of processing by hardware, but it is not necessary to be limited to such a configuration. For example, a configuration for processing by software is also possible. In this case, a general-purpose or dedicated computer equipped with a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), etc. can realize all or part of the above processing by the CPU. By reading the program from a computer-readable recording medium in which the program is recorded, developing the program in a ROM, a RAM, etc., and executing information processing / arithmetic processing, the same processing as the above-described image processing apparatus is performed. make it happen.

  Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

DESCRIPTION OF SYMBOLS 11 Processing area | region setting part 12 Block extraction part 13 Similarity calculation part 14 Weighting part 15 Flatness index value calculation part 18 Color component determination part

Claims (10)

A processing region setting means for determining a processing region that is a processing target region of the input image;
In the processing region, a block extracting unit that extracts a region centered on the target pixel as a target block and extracts a region having the same size as the target block as a reference block around a reference pixel different from the target pixel;
Similarity calculation means for calculating the similarity between the target pixel and the reference pixel based on a difference in pixel value between a pixel included in the target block and a pixel included in the reference block;
A weighting unit that calculates a lower weight value for the reference pixel according to the similarity, the higher the similarity is, and the lower the similarity is,
A flatness index value calculating means for calculating a flatness index value as an index of flatness of the processing area based on the pixel value of each pixel in the processing area and the weight value for the reference pixel;
An image processing apparatus comprising:   The flatness index calculation means includes a feature quantity calculation means for calculating a feature quantity corresponding to the amplitude of a high frequency component of each pixel in the processing region, and weights the feature quantity according to the weight value so that the flatness index The image processing apparatus according to claim 1, wherein the value is calculated.   The flatness index value calculating means calculates the feature amount based on a difference in pixel values between adjacent pixels in each of a plurality of predetermined directions in the processing region, and calculates a weighted average of the feature amount according to the weight value. The image processing apparatus according to claim 2, wherein the image processing apparatus calculates and sets a maximum value of the weighted average for each predetermined direction as a flatness index value.   The feature amount calculating means calculates a value based on a difference in pixel value between adjacent pixels in each of a plurality of predetermined directions for each pixel in the processing region, and The image processing apparatus according to claim 2, wherein a maximum value based on the difference is a feature amount of each pixel.   The weighting means sets the weight value to 1 when the similarity index value representing the similarity is lower than a predetermined threshold, and sets the weight value to zero when the similarity index value is higher than the threshold. The image processing apparatus according to any one of claims 1 to 4.   The weighting means calculates edge information indicating the degree of edge in each pixel of the input image, and the weight value decreases as the similarity between the edge information of the target pixel and the edge information of the reference pixel decreases. The image processing apparatus according to claim 1, wherein the weight value is corrected.   The weighting unit calculates edge information indicating a degree of edge in each pixel of the input image, and corrects the weight value so that the weight value becomes low when the value of edge information of a reference pixel is high. The image processing apparatus according to any one of claims 1 to 6. When the input image is an image signal lacking a specific color component, color component determination means for determining a color component in each pixel of the processing region and outputting it as color information,
The similarity calculation means calculates, for each color information, a difference between a pixel value of a pixel included in the target block and a pixel included in the reference block, and based on the difference, between the target pixel and the reference pixel The image processing apparatus according to claim 1, wherein the similarity is calculated. A processing region setting step for determining a processing region which is a processing target region in the input image;
In the processing region, a block extracting step of extracting a region centered on the target pixel as a target block and extracting a region having the same size as the target block as a reference block around a reference pixel different from the target pixel;
A similarity calculation step of calculating a similarity between the target pixel and the reference pixel based on a difference in pixel value between a pixel included in the target block and a pixel included in the reference block;
A weighting step for calculating a weight value that is higher as the similarity is higher and lower as the similarity is lower, according to the similarity, with respect to the reference pixel;
A flatness index value calculating step for calculating a flatness index value serving as an index of flatness of the processing region based on a pixel value of each pixel in the processing region and the weight value for the reference pixel;
An image processing method comprising: A processing region setting step for determining a processing region which is a processing target region in the input image;
In the processing region, a block extracting step of extracting a region centered on the target pixel as a target block and extracting a region having the same size as the target block as a reference block around a reference pixel different from the target pixel;
A similarity calculation step of calculating a similarity between the target pixel and the reference pixel based on a difference in pixel value between a pixel included in the target block and a pixel included in the reference block;
A weighting step for calculating a weight value that is higher as the similarity is higher and lower as the similarity is lower, according to the similarity, with respect to the reference pixel;
A flatness index value calculating step for calculating a flatness index value serving as an index of flatness of the processing region based on a pixel value of each pixel in the processing region and the weight value for the reference pixel;
An image processing program for causing a computer to execute.

Images