JP2010246053A - Apparatus, method, and program for processing image, and printer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make only a truly necessary corrections to be performed by correctly specifying a red eye pixel as a pixel to be corrected. <P>SOLUTION: An image processing apparatus includes: a region detection part for detecting a region to be corrected that includes a red eye region from within an image represented by image data obtained; a pixel identification part for calculating a prescribed character value for every pixel included in the region to be corrected that is detected and identifying a pixel to be an object of red eyes correction among pixels that are included in the region to be corrected in accordance with the characteristic value; and a red eye correction part for correcting red eye with the pixel identified as an object. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, an image processing method, an image processing program, and a printing apparatus.

  When the person's eyes are so-called “red eyes” in an image obtained by flash photography of the person, correction is applied to the image to alleviate the state of red eyes. Until now, it is determined by comparing the color information for each pixel with a predetermined threshold value whether any pixel in the image is a pixel corresponding to the red eye (red-eye pixel). The above correction is performed on the pixel determined to be a pixel.

  In addition, a region including a part of the subject's head that may have one or more pupil color defects may be determined, pixels in the region may be sampled, and each of the sampled pixels may represent a pupil color defect A method is known in which a pixel having a possibility of poor pupil color tone is identified in comparison with the threshold value (see Patent Document 1).

Japanese Patent No. 3181472

  In many cases, the red eye pixel is quite similar to other pixels (for example, skin color pixels) whose color information (or part of the color information) is not a red eye pixel. Therefore, in the simple comparison determination between the color information and the threshold as described above, pixels that are not actually red-eye pixels are often determined to be red-eye pixels because the colors are close to the red-eye pixels. In some cases, the correction is performed even for pixels that do not require correction for reducing the red-eye state.

  The present invention has been made in view of the above problems, and by accurately specifying a pixel corresponding to red eye as a correction target, only a truly necessary correction can be performed, and an image with excellent image quality can be obtained. An object of the present invention is to provide an image processing apparatus, an image processing method, an image processing program, and a printing apparatus.

In order to achieve the above object, an image processing apparatus according to the present invention includes an area detection unit that detects a correction candidate area including a red-eye area from an image represented by acquired image data, and pixels included in the detected correction candidate area. A predetermined feature value is calculated for each pixel, and a pixel specifying unit that specifies a pixel to be subjected to red-eye correction among pixels included in the correction candidate region according to the feature value, and the specified pixel as a target A red-eye correction unit that performs the red-eye correction.
According to the present invention, the target pixel for red-eye correction is specified from the pixels included in the correction candidate area according to the feature value, and the red-eye correction is performed on the specified pixel. As a result, red-eye correction is performed only on (pixels), and unnecessary correction is eliminated.

The pixel specifying unit calculates a feature value based on an inverted value of the R value of the RGB values representing the pixel, and specifies the pixel according to a comparison result between the feature value and a predetermined threshold value. Also good. According to this configuration, since the feature value is calculated based on the inverted value of the R value in which the feature as the red-eye pixel appears, the red-eye pixel can be accurately identified according to the feature value.
More specifically, the pixel specifying unit uses a value obtained by adding a gray value of a pixel to the inversion value as a feature value. Alternatively, the pixel specifying unit adds the inverted value to each of the RGB values, and uses a luminance value calculated based on RGB after adding the inverted value as a feature value. Since such feature values differ greatly between red-eye pixels and pixels other than red-eye pixels, the red-eye pixels can be accurately identified by comparing such feature values with threshold values.

  The pixel specifying unit and the red eye correcting unit may read the image data in a raster unit, and repeat the specification of the pixel and the red eye correction to the specified pixel for each image data in the raster unit. According to this configuration, the pixel specification and the red-eye correction can be accurately performed even in an environment (for example, printer firmware) in which the memory area used for holding the image data is small when the pixel is specified and the red-eye correction is performed. it can.

  The technical idea of the present invention can be realized by means other than an image processing apparatus. For example, an invention of a method including processing steps executed by each unit of an image processing apparatus, or an invention of an image processing program that causes a device (printing device, computer, imaging device, etc.) to execute a process executed by each unit of the image processing device Can also be grasped. Further, an area detection unit that detects a correction candidate area including a red-eye area from the image represented by the acquired image data, calculates a predetermined feature value for each pixel included in the detected correction candidate area, and the feature A pixel identifying unit that identifies a pixel that is a target of red-eye correction among pixels included in the correction candidate region according to a value; a red-eye correcting unit that performs the red-eye correction on the identified pixel; and the red-eye A printing apparatus that includes a printing unit that performs printing based on the image data after the correction is performed can also be regarded as one invention.

FIG. 2 is a block diagram schematically illustrating a configuration of a printer. 6 is a flowchart illustrating image correction processing executed by a printer. It is the figure which showed an example of the detection result of the correction | amendment candidate area | region containing a red-eye area | region. It is the flowchart which showed an example of the detail of preprocessing. It is the flowchart which showed the other example of the detail of pre-processing. It is the figure which showed R value of the pixel on the horizontal line which passes a red-eye area | region, and R value of the pixel on the horizontal line which passes a skin area | region. It is the figure which showed the feature value of the pixel on the horizontal line which passes a red-eye area | region, and the feature value of the pixel on the horizontal line which passes a skin area | region. It is the figure which showed the luminance value of the pixel on the horizontal line which passes a red-eye area | region, and the luminance value of the pixel on the horizontal line which passes a skin area | region. It is the figure which showed the feature value of the pixel on the horizontal line which passes a red-eye area | region, and the feature value of the pixel on the horizontal line which passes a skin area | region. 5 is a flowchart showing details of red-eye correction. It is the figure which showed an example of the histogram for every RGB. It is the figure which showed an example of the correction coefficient determination table. It is the figure which showed an example of the distance coefficient determination table.

1. FIG. 1 schematically shows the configuration of a printer 10 corresponding to an example of an image processing apparatus and a printing apparatus of the present invention. The printer 10 includes a CPU 11, an internal memory 12 including a ROM and a RAM, a general-purpose interface (GIF) 13, an operation unit 14 configured with buttons and a touch panel, a display unit 15 configured with a liquid crystal display, and a print engine. 16. Each of these elements 11 to 16 realizes communication by a bus 17. The CPU 11 expands a predetermined program (such as an image processing program) stored in the ROM to the RAM and performs arithmetic processing according to the program.

  The GIF 13 provides an interface conforming to the USB standard, for example, and can be connected to a digital still camera (DSC) 20 or a computer 30. The printer 10 can input image data representing an image to be printed from the DSC 20 or the computer 30 connected via the GIF 13. The GIF 13 also has a function of a card interface connected to a card slot (not shown), and inputs image data stored in the memory card 40 from a predetermined storage medium (memory card) 40 inserted in the card slot. be able to.

  The print engine 16 is a printing mechanism that performs printing based on print data, and employs, for example, an inkjet printing mechanism. Of course, the printer 10 is not limited to the ink jet type, and may be a laser printer, an LED printer, or the like. The printer 10 may be a printing device for consumers, a business printing device for DPE (a so-called minilab machine), or various functions such as a copy function and a scanner function in addition to the print function. A so-called multifunction machine having the above functions may be used.

  FIG. 1 shows an image correction unit 121, a display processing unit 122, a print processing unit 123, and the like as functional blocks realized by a program in the internal memory 12. The image correction unit 121 includes a region detection unit 121a, a pixel specifying unit 121b, and a red-eye correction unit 121c as program modules. The functions of these units will be described later. The display processing unit 122 is a display driver that controls the display unit 15 to display a user interface (UI), a message, a thumbnail image, or the like on the screen of the display unit 15. The print processing unit 123 performs printing by appropriately performing predetermined processing such as color conversion processing and halftone processing on image data to be printed (for example, image data after image correction performed by the image correction unit 121). This is a printer driver for generating data and driving the print engine 16 to print an image based on the print data.

2. Image Correction Processing FIG. 2 is a flowchart showing image correction processing executed by the printer 10. This process is mainly realized by the function of the image correction unit 121.
In step S (hereinafter, step notation is omitted) 100, the image correction unit 121 acquires image data representing one image to be processed from the DSC 20, the computer 30, the memory card 40, and the like. Read into the internal memory 12. The image data expresses an image by a plurality of pixels, and each pixel is expressed by a combination of gradations (for example, 256 gradations of 0 to 255) of RGB (red, green, blue) channels. . The image data may be compressed when recorded in the memory card 40 or the like, or the color of each pixel may be expressed in another color system. In these cases, the image correction unit 121 executes image data expansion and color system conversion to acquire image data in a required color system. In the present embodiment, the image data acquired in S100 is a photographic image including at least a person's face (especially eyes), and the person's eyes are in a “red-eye” state due to flash photography. Suppose that

  In S110, the area detection unit 121a detects a correction candidate area A2 including the red-eye area A1 from the image represented by the image data acquired in S100. In this case, the area detection unit 121a first detects the red-eye area A1 from the image. The red eye region is a generally circular image region that matches the characteristic color pattern of the red eye. The region detection unit 121a detects a red-eye region by a known method (for example, a method described in Japanese Patent Application Laid-Open No. 2007-156694). After detecting the red-eye area A1, the area detection unit 121a sets a correction candidate area A2 including the red-eye area A1 on the image data. For example, the area detection unit 121a sets a rectangle including the entire red eye area A1 on the image data, and sets the rectangle as the correction candidate area A2.

  FIG. 3 is a diagram illustrating an example of the detection result in S110. In the example of FIG. 3, the person P1 is included in the image data D, and a part of the left and right eyes of the person P1 are detected as red-eye areas A1 and A1, respectively. Further, correction candidate areas A2 and A2 are set for each of the red-eye areas A1 and A1. For example, the ratio of the size of the correction candidate area A2 to the red eye area A1 is determined in advance, and the correction candidate area A2 is set so that the center of gravity of the red eye area A1 substantially coincides with the center of gravity. The area detection unit 121a acquires position information (for example, coordinates of the upper left vertex of the rectangle) and size information (vertical width and horizontal width) for each correction candidate area A2 on the image data as a detection result of the correction candidate area A2. save. The shape of the correction candidate area A2 is not limited to a rectangle, but may be a circle.

In S <b> 120, the image correction unit 121 selects one raster among the rasters constituting the image data acquired in S <b> 100 as a processing target raster and stores it in a predetermined buffer in the internal memory 12. A raster is a pixel row in the horizontal direction (x direction) in image data.
In S <b> 130, the image correction unit 121 sets one pixel among the pixels constituting the processing target raster selected in the latest S <b> 120 as the target pixel.

  In S140, the pixel specifying unit 121b determines whether or not the target pixel set in the latest S130 belongs to the correction candidate area A2 detected in S110, and if the target pixel belongs to any correction candidate area A2. If it is determined, the process proceeds to S150, and if it is determined that the target pixel does not belong to any correction candidate area A2, the process proceeds to S180. Whether or not the pixel of interest belongs to the correction candidate region A2 is determined based on the coordinates of the pixel of interest on the image data and the position information and size information stored for each correction candidate region A2 in S110.

  In S120, one raster to be processed is selected. However, depending on the raster, none of the pixels constituting the raster may belong to the correction candidate area A2. For such a raster, it is useless to repeat the determination of S140 for each pixel. Therefore, when the pixel specifying unit 121b performs the determination in S140 for the first target pixel in the processing target raster selected in S120, the pixel specifying unit coordinates in the vertical direction (y direction) in the image data and the y direction Is compared with the existence range of the correction candidate area A2, and if the target pixel does not belong to such an existence range, it is determined that other pixels belonging to the common raster do not belong to the correction candidate area A2, and the process proceeds to S190. It is good also as advancing (dashed line in FIG. 2).

  In S150, the pixel specifying unit 121b performs predetermined preprocessing on the target pixel set in the latest S130. Pre-processing is processing performed to further increase the accuracy of the determination result prior to determining whether the target pixel is a pixel corresponding to red eyes (red-eye pixel) (S160). As the preprocessing, the pixel specifying unit 121b calculates a feature value based on the inverted value of the R value of the RGB values representing the target pixel.

FIG. 4 is a flowchart showing an example of details of the preprocessing.
In S151, the pixel specifying unit 121b obtains an inverted value Rre of the R value of the target pixel. The inversion value Rre = 255−R. In S152, the pixel specifying unit 121b obtains the calculated value Rc by adding the gray value J of the target pixel to the inverted value Rre. The gray value may simply be a luminance value calculated based on RGB of the target pixel, or may be a gray value calculated by a predetermined photo retouching application for the target pixel. The luminance value can be calculated by known weighted addition of RGB components. In S153, the pixel specifying unit 121b stores the calculated value Rc (= Rre + J) thus obtained as the feature value of the target pixel.

FIG. 5 is a flowchart showing an example of details of pre-processing different from FIG.
In S155, the pixel specifying unit 121b obtains an inverted value Rre of the R value of the target pixel. In S156, the pixel specifying unit 121b adds the inverted value Rre to each of RGB of the target pixel. That is, the operation values (R + Rre, G + Rre, B + Rre) are obtained. In S157, the pixel specifying unit 121b regards the three components (R + Rre, G + Rre, B + Rre) as the calculated values as RGB components, calculates Yc as the luminance value by the above weighted addition, and pays attention to the luminance value Yc. Save as pixel feature values.

  In S160, the pixel specifying unit 121b determines whether the target pixel is a red-eye pixel according to the feature value calculated in the latest S150. That is, the pixel specifying unit 121b specifies a pixel that is a target of red-eye correction (S170) described later. Specifically, when the calculated value Rc is calculated as the feature value in S150, the pixel specifying unit 121b is stored in advance in the internal memory 12 or the like for comparison between the calculated value Rc and the calculated value Rc. The threshold value TH1 is compared, and when the calculated value Rc is equal to or less than the threshold value TH1, it is determined that the target pixel is a red-eye pixel (Yes), and the process proceeds to S170. In addition, when the pixel specifying unit 121b calculates the luminance value Yc as the feature value in S150, the threshold value stored in the internal memory 12 or the like in advance for comparison between the luminance value Yc and the luminance value Yc. When the luminance value Yc is equal to or less than the threshold value TH2, it is determined that the pixel of interest is a red-eye pixel (Yes), and the process proceeds to S170. On the other hand, when the feature value exceeds the threshold value to be compared, the pixel specifying unit 121b determines “No”, skips S170, and proceeds to determination of S180.

  6 and 7 exemplify the R value and the feature value (calculated value Rc) for each pixel in the correction candidate area A2 in certain image data. In FIG. 6, the R value (solid line) for each pixel on the horizontal line H1 passing through the red-eye area A1 in the correction candidate area A2 and the horizontal line H2 passing through the skin area below the eyes in the correction candidate area A2 The R value (dashed line) for each pixel is shown in the graph. On the other hand, in FIG. 7, the calculated value Rc (solid line) for each pixel on the same horizontal line H1 and the calculated value Rc (one-dot chain line) for each pixel on the horizontal line H2 are shown in a graph. 6 and 7, the horizontal axis indicates the pixel position in the horizontal direction in the correction candidate area A2, and the vertical axis indicates the gradation value. 6 and 7, the range corresponding to the red eye (the range excluding the highlight portion) in the pupil of the eyeball is surrounded by a chain line rectangle.

  As can be seen from FIG. 6, the R value of the pixel corresponding to the red eye among the pixels on the horizontal line H1 is somewhat high, but the R value of the pixel corresponding to the skin on the horizontal line H2 is also high. Therefore, even if the R value of each pixel in the correction candidate area A2 is separated with a simple threshold, the red-eye pixel and the pixel other than the red-eye pixel cannot be accurately separated. On the other hand, as can be seen from FIG. 7, the calculated value Rc of the pixel corresponding to the red eye protrudes considerably lower than the calculated value Rc of the pixel corresponding to other black eyes, white eyes, and skin areas. Therefore, if the threshold value TH1 is set to the gradation level as indicated by the two-dot chain line in FIG. 7, it is possible to accurately determine whether or not the target pixel is a red-eye pixel in S160. it can.

  8 and 9 illustrate the luminance value Y and the characteristic value (luminance value Yc) for each pixel in the correction candidate area A2 in certain image data. In FIG. 8, the luminance value Y (solid line) for each pixel on the horizontal line H1 that passes through the red-eye area A1 in the correction candidate area A2 and the horizontal line H2 that passes through the skin area below the eyes in the correction candidate area A2 The luminance value Y (one-dot chain line) for each of the upper pixels is shown in the graph. On the other hand, in FIG. 9, the luminance value Yc (solid line) for each pixel on the same horizontal line H1 and the luminance value Yc (one-dot chain line) for each pixel on the horizontal line H2 are shown in a graph. 8 and 9, as in FIGS. 6 and 7, the horizontal axis indicates the horizontal pixel position in the correction candidate area A2, the vertical axis indicates the gradation value, and the red eye (highlighted portion) in the pupil of the eyeball. The range corresponding to (excluding the range) is surrounded by a chain-line rectangle.

  As can be seen from FIG. 8, the luminance value Y takes a low value not only in the red-eye range but also in the pixel corresponding to the black-eye range among the pixels on the horizontal line H1. Therefore, even if the luminance value Y of each pixel in the correction candidate area A2 is separated by a simple threshold value (for example, the threshold value TH3 shown in FIG. 8), the red-eye pixel and the pixels other than the red-eye pixel are accurately distinguished. Cannot be divided into. On the other hand, as can be seen from FIG. 9, the luminance value Yc of the pixel corresponding to the red eye protrudes considerably downward as compared with the luminance value Yc of the pixel corresponding to the other black eyes, white eyes, and skin regions. Therefore, if the threshold value TH2 is set to the gradation level as indicated by the two-dot chain line in FIG. 9, it is possible to accurately determine whether or not the target pixel is a red-eye pixel in S160. it can.

  The pixel specifying unit 121b calculates the calculated value Rc by repeating S150 and S160 for all the pixels in the correction candidate area A2 and compares the calculated value Rc with the threshold value TH1, or for all the pixels, S150 and S160. Is repeated to calculate the luminance value Yc and compare it with the threshold value TH2. However, the image specifying unit 121b calculates the calculated value Rc and the luminance value Yc for all the pixels, and determines the pixels whose calculated value Rc is equal to or lower than the threshold value TH1 and whose luminance value Yc is equal to or lower than the threshold value TH2. You may specify as an object of red eye correction (S170).

  In S170, the red-eye correction unit 121c performs red-eye correction on the target pixel identified (determined) as a red-eye pixel in the latest S160. In general, the red-eye correction calculates the saturation of the target pixel, determines the correction degree of the target pixel according to the calculated saturation, and the color information (particularly, the R value) is determined based on the determined correction degree. This refers to a process of correcting the target pixel so as to decrease.

FIG. 10 is a flowchart showing an example of details of red-eye correction.
In S171, the red-eye correction unit 121c corrects the RGB value of the target pixel by adjusting the color balance in the image area including the target pixel. The color balance adjustment is a process for reducing the deviation of the RGB distribution in the image area. By performing the color balance adjustment, for example, when the image has a red fog, the red fog can be reduced. . The method of color balance adjustment is not particularly limited. For example, the red-eye correction unit 121c uses a histogram for each RGB by using all the pixels constituting the raster to which the pixel of interest belongs (the processing target raster selected in the latest S120). (Frequency distribution) is generated.

  FIG. 11 illustrates a histogram for each RGB generated in the process of color balance adjustment in S171. The red-eye correction unit 121c calculates a value on the lower gradation side by the number of pixels corresponding to a predetermined percentage (for example, 10%) of the pixels used for generating the histogram from the maximum values Rmax, Gmax, and Bmax of the RGB histograms. These are recognized as adjustment reference values Rs, Gs, and Bs. Then, the color balance adjustment amount for each of RGB is obtained based on the relative gradation difference (for example, ΔGR = Gs−Rs, ΔGB = Gs−Bs) between the components of the adjustment reference value RsGsBs, and based on the adjustment amount. The RGB value of each pixel used for generating the histogram is corrected. As a result, the RGB value of each pixel constituting the raster to which the pixel of interest belongs as well as the pixel of interest is corrected, and the color balance in the raster is adjusted. The red-eye correction unit 121c stores the RGB values (corrected R′G′B ′) after color balance adjustment (corrected) for each pixel.

  In the present embodiment, red-eye correction is not performed on pixels that do not belong to at least the correction candidate area A2. Therefore, in S171, the correction based on the adjustment amount may be performed only for the pixels belonging to the correction candidate area A2 among the pixels constituting the raster to which the target pixel belongs. The red-eye correction flowchart of FIG. 10 is basically repeatedly executed for each pixel of interest, but the color balance adjustment in S171 is performed collectively for each pixel having a common raster. For this reason, after performing color balance adjustment for the processing target raster to which the target pixel belongs in S170 (S171) after the one target pixel is determined to be a red-eye pixel in S160, other target pixels belonging to the same processing target raster are included. Even if the pixel is the pixel of interest and is determined to be a red-eye pixel in S160, the color balance adjustment in S171 is not performed, and the result of the color balance adjustment that has already been performed (the pixel of interest at that time has already been saved. The process after S172 is performed using the corrected R′G′B ′).

In S172, the red-eye correction unit 121c calculates the luminance value Y ′ by the weighted addition based on the corrected R′G′B ′ of the target pixel, and the saturation based on the corrected R′G′B ′. S is calculated. Saturation S can be obtained from saturation S = (Imax−Imin) · 255 / Imax, where Imax is the maximum value and Imin is the minimum value of each component of R′G′B ′ after correction.
In S173, the red-eye correction unit 121c calculates the correction degree according to the saturation S.

  FIG. 12 illustrates the correction coefficient determination table T1 used in S173. The correction coefficient determination table T1 is a function in which the horizontal axis represents saturation S and the vertical axis represents correction coefficient (0.0 to 1.0) Cs. The correction coefficient Cs = 1.0 means that the correction degree is minimum (not corrected), and the correction degree increases as the correction coefficient Cs decreases. In this embodiment, the correction coefficient determination table T1 is stored in advance in the internal memory 12 (see FIG. 1), and the red-eye correction unit 121c refers to the correction coefficient determination table T1 to correct according to the saturation S. The coefficient Cs (correction degree) is determined. The correction coefficient determination table T1 has a characteristic that the correction coefficient Cs decreases as the input value (saturation S) increases. Specifically, when the input value (saturation S) exceeds a predetermined saturation Sr, the correction coefficient that has been high until then is obtained. It has the characteristic that Cs decreases rapidly. The saturation Sr corresponds to, for example, the minimum saturation value of red-eye pixels obtained in advance through experiments or the like.

In S174, the red-eye correction unit 121c corrects the corrected R ′ among the components of the corrected R′G′B ′ according to the determined correction degree. That is, the corrected R ″ is calculated by multiplying the corrected R ′ by the determined correction coefficient Cs (R ′ × Cs).
In S175, the red-eye correction unit 121c corrects the values of G and B so that the ratio between the components of the target pixel is maintained before and after the correction. Specifically, the corrected G ″ and the corrected B ″ are calculated based on the ratio between the corrected R′G′B ′ components and the corrected R ″.
That is, the corrected G ″ and the corrected B ″ that satisfy R ′: G ′ = R ″: G ″ and R ′: B ′ = R ″: B ″ are calculated.

In S176, the red-eye correction unit 121c calculates the luminance value Y ″ (corrected luminance value Y ″) by the above weighted addition based on the corrected R ″ G ″ B ″ and the attention before and after the correction. Pixel luminance difference ΔY = Y′−Y ″ is calculated.
In S177, the red-eye correction unit 121c corrects the target pixel so that the luminance difference between the target pixel before and after the correction is substantially smaller as the distance from the center of gravity of the correction candidate area A2 is longer. Here, it is assumed that the center of gravity of the correction candidate area A2 and the center of gravity of the red-eye area A1 included therein match, but if the centers of gravity do not match, the red-eye correction unit 121c determines that the luminance difference is The pixel of interest may be corrected so that the pixel farther from the center of gravity of the red-eye region A1 becomes substantially smaller.

  FIG. 13 illustrates the distance coefficient determination table T2 used in S177. The distance coefficient determination table T2 is a function in which the horizontal axis is the distance d of the pixel of interest from the center of gravity of the correction candidate area A2, and the vertical axis is the distance coefficient Cd (where Cd ≧ 0). In the present embodiment, the distance coefficient determination table T2 is stored in the internal memory 12 in advance (see FIG. 1), and the red-eye correction unit 121c refers to the distance coefficient determination table T2 to thereby determine the distance coefficient corresponding to the distance d. Cd is determined. The distance coefficient determination table T2 has a characteristic that the distance coefficient Cd increases as the input value (distance d) increases.

  The red-eye correction unit 121c calculates a corrected luminance difference ΔY ′ by multiplying the luminance difference ΔY by the determined distance coefficient Cd (ΔY × Cd). Then, the corrected luminance difference ΔY ′ is added to each corrected R ″ G ″ B ″ of the target pixel. That is, the red-eye correction unit 121c obtains Rf = R ″ + ΔY ′, Gf = G ″ + ΔY ′, and Bf = B ″ + ΔY ′ as calculation results, and uses these RfGfBf as the final red-eye correction for the target pixel. Save as a good result. As described above, the distance coefficient Cd that determines the corrected luminance difference ΔY ′ added to each of the corrected R ″ G ″ B ″ becomes larger as the pixel of interest with a longer distance d. Therefore, the pixel of interest represented by RfGfBf basically has a smaller variation in luminance before and after red-eye correction as the distance d is longer.

Returning to FIG. 2, the description will be continued.
In S180, the image correction unit 121 determines whether or not all the pixels constituting the processing target raster selected in the latest S120 have been set as the target pixel, and sets all the pixels of the processing target raster as the target pixel. If it is determined that the setting has been completed, the process proceeds to S190. On the other hand, if there is a pixel that has not been set as the target pixel among the pixels of the processing target raster, the process returns to S130, and one of the unset pixels is set as the target pixel, and the processes from S140 onward are performed. repeat.

  In S190, the image correction unit 121 determines whether or not all the rasters constituting the image data acquired in S100 have been selected as the processing target rasters, and has completely selected all the rasters as the processing target rasters. 2 is finished, the flowchart of FIG. On the other hand, if there is a raster that is not selected as the processing target raster among all the rasters, the process returns to S120, and one of the unselected rasters is selected as the processing target raster, and the processing from S130 onward is performed. repeat. As a result, the red-eye correction is executed only for the pixels constituting the image data that belong to the correction candidate area A2 and are determined to correspond to the red-eye pixel by comparing the feature value with the threshold value. .

3. Summary As described above, according to the present embodiment, for each pixel belonging to the correction candidate area A2 detected as the area including the red eye from the image data, only the pixel having a high red-eye property corresponds to the other black eye part, the white eye part, and the skin part. The feature value is calculated by performing preprocessing to change the value to a pixel that is significantly different from the pixel to be corrected, and based on the comparison between the feature value and a predetermined threshold value, each pixel belonging to the correction candidate area A2 becomes a red-eye pixel. It is determined whether or not it corresponds, and the red-eye correction is executed only for the pixel that is determined to be a red-eye pixel. In other words, the determination as to whether or not the red-eye pixel is less accurate in the conventional simple comparison between the pixel value of the image data and the threshold value is made extremely accurate. As a result, it is possible to prevent only the pixels that actually correspond to the red eye on the photographed image from being subjected to the red eye correction and to apply the red eye correction to pixels that originally do not require the red eye correction. In addition, when the printer 10 performs printing based on the image data after the red-eye correction is performed on the red-eye pixel in this way, it is possible to obtain a human image with the red-eye appropriately corrected as a print result.

  In red-eye correction, attention is paid to the fact that red-eye pixels have high saturation. By increasing the degree of correction as the saturation increases, the R value is greatly reduced for pixels with higher saturation. Therefore, only the red-eye pixel appropriately reduces its redness, and an image in which red-eye is eliminated can be obtained. In particular, the degree of correction increases rapidly when the saturation of the pixel becomes equal to or higher than the saturation Sr (the correction coefficient Cs decreases rapidly), so that the redness of the pixel that truly corresponds to the red eye can be appropriately reduced. it can. Further, since the red balance of the image is corrected by adjusting the color balance before determining the correction degree according to the saturation of the pixel, the calculated saturation is not an extremely high value. As a result, overcorrection is prevented in the correction for reducing the R value of the pixel.

  Furthermore, since the correction is performed so that the ratio of the RGB components of the correction target pixel is substantially maintained before and after the red-eye correction, the color of the correction target pixel is prevented from greatly changing before and after the correction. Furthermore, in red-eye correction, the pixel farther from the center of gravity of the correction candidate area A2 is corrected so that the luminance difference before and after correction becomes smaller. Therefore, the luminance changes abruptly at the boundary between the corrected area and the uncorrected area. Thus, it is possible to prevent the image from being uncomfortable. Furthermore, since the image correction unit 121 performs preprocessing, determination of whether or not it is a red-eye pixel, and red-eye correction based on the raster unit image data, each of these processes can be performed with less risk, and particularly for an image before printing. This is a preferable configuration in the printer 10 in which there is a limit to resources that can be used for correction processing and the like.

  In the above description, the printer 10 has been described as an execution subject of the present embodiment. However, hardware other than the printer 10, for example, the DSC 20 or the computer 30 is provided with the function of the image correction unit 121. It is good also as a structure which performs the flowchart of 2. FIG.

DESCRIPTION OF SYMBOLS 10 ... Printer, 11 ... CPU, 12 ... Internal memory, 16 ... Print engine, 20 ... DSC, 30 ... Computer, 40 ... Memory card, 121 ... Image correction part, 121a ... Area | region detection part, 121b ... Pixel specific part, 121c ... Red-eye correction unit, 122 ... Display processing unit, 123 ... Print processing unit

Claims (8)

An area detection unit that detects a correction candidate area including a red-eye area from the image represented by the acquired image data;
A pixel specification for calculating a predetermined feature value for each pixel included in the detected correction candidate area and specifying a pixel to be subjected to red-eye correction among the pixels included in the correction candidate area according to the feature value And
An image processing apparatus comprising: a red-eye correction unit that performs the red-eye correction on the identified pixel.   The pixel specifying unit calculates a feature value based on an inverted value of an R value of RGB values representing the pixel, and specifies the pixel according to a comparison result between the feature value and a predetermined threshold value. The image processing apparatus according to claim 1.   The image processing apparatus according to claim 2, wherein the pixel specifying unit uses a value obtained by adding a gray value of a pixel to the inversion value as a feature value.   3. The image according to claim 2, wherein the pixel specifying unit adds the inverted value to each of the RGB values and uses a luminance value calculated based on RGB after adding the inverted value as a feature value. 4. Processing equipment.   The pixel specifying unit and the red-eye correcting unit read image data in raster units, and repeat the specification of the pixels and execution of red-eye correction on the specified pixels for each image data in raster units. The image processing apparatus according to claim 1. An area detection step of detecting a correction candidate area including a red-eye area from the image represented by the acquired image data;
A pixel specification for calculating a predetermined feature value for each pixel included in the detected correction candidate area and specifying a pixel to be subjected to red-eye correction among the pixels included in the correction candidate area according to the feature value Process,
And a red-eye correction step of performing the red-eye correction on the identified pixel. An area detection function for detecting a correction candidate area including a red-eye area from the image represented by the acquired image data;
A pixel specification for calculating a predetermined feature value for each pixel included in the detected correction candidate area and specifying a pixel to be subjected to red-eye correction among the pixels included in the correction candidate area according to the feature value Function and
An image processing program for causing a computer to execute a red-eye correction function for performing the red-eye correction on the identified pixel. An area detection unit that detects a correction candidate area including a red-eye area from the image represented by the acquired image data;
A pixel specification for calculating a predetermined feature value for each pixel included in the detected correction candidate area and specifying a pixel to be subjected to red-eye correction among the pixels included in the correction candidate area according to the feature value And
A red-eye correction unit that performs the red-eye correction on the identified pixel;
And a printing unit that executes printing based on the image data after the red-eye correction is performed.

Images