JP2004297439A - Image correction method, apparatus, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image correction method, apparatus, and program capable of properly correcting the density or the like of an image whose dynamic range is extremely small. <P>SOLUTION: An image correction processing includes: a step 100 of interleaving image data of a processing object in units of pixels to generate reduced image data; a step 102 of generating histograms of R, G, B, Y, Cb and Cr; a step 104 of thereafter obtaining a color balance correction amount; steps 106 to 120 of extracting a principal region in the image to obtain a density correction condition; a step 122 of obtaining a saturation correction amount; a step 124 of setting a first correction condition; a step 126 of extracting the density of a highlight point from the histograms of R, G, B and extracting a gradation value at a shadow point from the histogram of Y; a step 130 of setting a second correction condition from them; a step 132 of obtaining the dynamic range of the image; a step 134 of using a weight correction set in response to the dynamic range to generate conversion data equivalent to the weighted average between the first correction condition and the second correction condition and converting the image data of the processing object by means of a 3D-LUT by using the generated conversion data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image correction apparatus, method, and program, and more particularly to an image correction method for correcting image density and color balance, an image correction apparatus to which the image correction method can be applied, and a computer. Related to the program to function as.
[0002]
[Prior art]
Various correction methods have been proposed as correction methods for correcting the density and color balance of an image obtained by shooting an arbitrary subject using a camera or a digital still camera (DSC). For example, the average density of the standard density region is calculated from the correction target image, and the average density and area ratio of the main image portion extracted from the correction target image are calculated. Based on the calculation result, the correction target image is calculated. A technique has been proposed in which a density correction curve is determined so that the calculated representative density becomes a target density, and the density of the image to be corrected is corrected based on the density correction curve (for example, Patent Documents). 1).
[0003]
[Patent Document 1]
JP 2002-199221 A
[0004]
[Problems to be solved by the invention]
The above technique is based on the fact that it is almost always a fixed part (main image part) that is noticed in image viewing, for example, a person's face is most noticeable in a portrait. However, by applying the above technique, it is possible to properly correct the density and the like for the majority of images. However, the above technique uses a dynamic range (maximum There is a drawback that it is difficult to appropriately correct the density of an image having a very small difference between the density and the minimum density.
[0005]
Recently, a request form for requesting photo processing such as creation of a photo print via the Internet has become widespread, but in this request form, as in the case where simultaneous printing of images photographed and recorded on photographic film is requested, In many cases, the client who is the client himself / herself requests photo processing without confirming the contents of the image, and the images for which photo processing has been requested are mixed together as described above. It is also increasing. For this reason, establishment of a correction method capable of appropriately correcting the density and the like for an image having a very small dynamic range has been demanded.
[0006]
The present invention has been made in consideration of the above facts, and an object thereof is to obtain an image correction apparatus, an image correction method, and a program that can appropriately correct the density of an image having a very small dynamic range.
[0007]
[Means for Solving the Problems]
The inventor of the present application conducted an experiment to correct the density and the like by applying various known correction methods to an image having a very small dynamic range, and to evaluate the correction result. As a result, for an image having a very small dynamic range, the difference between the gradation value corresponding to the highlight of the image and the gradation value corresponding to the shadow (that is, the dynamic range) becomes a predetermined value (the original value). By applying a correction method that corrects the density etc. (so that it is enlarged than the image), the knowledge that the density etc. can be corrected properly by expanding the dynamic range of the image has been obtained, and the present invention has been achieved. It was.
[0008]
Based on the above, the image correction apparatus according to the first aspect of the present invention recognizes the main part region of the correction target image, and corrects the correction target image based on the recognized feature amount of the main part region. First correction means for calculating correction conditions and correcting the correction target image so that a difference between a gradation value corresponding to highlight and a gradation value corresponding to shadow in the correction target image becomes a predetermined value. Second calculating means for calculating the second correction condition, setting means for setting a weight for the first correction condition and the second correction condition based on a dynamic range of the correction target image, and the setting means Based on the weight set by the correction, the correction target image is corrected under the first correction condition calculated by the first calculation means, and the second correction condition is calculated by the second calculation means. Is configured to include a correction means for performing a weighted mean equivalent to correction of the correction, the.
[0009]
The first calculation means according to the first aspect of the invention recognizes a main part region of the correction target image, and corrects the correction target image based on the recognized feature amount of the main part region. Is calculated. Accordingly, it is possible to obtain a correction condition (first correction condition) that can appropriately correct the density and the like for the majority of images excluding a part of the images having a very small dynamic range. In addition, the second calculation means according to the first aspect of the invention is configured so that the difference between the gradation value corresponding to the highlight in the correction target image and the gradation value corresponding to the shadow becomes a predetermined value (for example, the floor). The second correction condition for correcting the correction target image is calculated (so that the difference between the tone values is enlarged from that of the original image). Thereby, it is possible to obtain a correction condition (second correction condition) that can appropriately correct the density and the like of an image having a very small dynamic range.
[0010]
According to the first aspect of the present invention, the weights for the first correction condition and the second correction condition are set by the setting means based on the dynamic range of the correction target image. For example, when the dynamic range of the correction target image is relatively large, the weight for the first correction condition is large, and when the dynamic range is very small, the weight for the second correction condition is large. Can be set to
[0011]
The correction means according to the first aspect of the invention is based on the weight set by the setting means and is corrected by the first correction condition calculated by the first calculation means and calculated by the second calculation means. Correction corresponding to the weighted average of correction under the second correction condition is performed on the correction target image. Thus, according to the first aspect of the present invention, the content of correction (the specific gravity of the correction under the first correction condition and the correction under the second correction condition) is corrected according to the dynamic range of the correction target image. Therefore, it is possible to appropriately correct the density of an image having a very small dynamic range.
[0012]
In the first aspect of the invention, the first calculating means is based on the distribution on the color coordinates of the pixel group whose saturation is equal to or less than a predetermined value in the correction target image. In addition to setting the color balance correction condition, the density correction condition is set so that the correction target density obtained from the feature amount of the recognized main area is converted to a predetermined target density. As the first correction condition, A correction condition for correcting the color balance and density of the correction target image under the set correction condition can be calculated. As a result, even when the correction target image is an image taken under illumination with a different type of light source, a correction condition for properly correcting the color balance of the correction target image can be obtained. It is possible to obtain a correction condition that can appropriately correct the density and color balance of the majority of images excluding some images with a very small range.
[0013]
Further, in the first aspect of the invention, as described in the third aspect of the invention, for example, the second calculation means has a gradation value corresponding to the highlight for each color component in the correction target image as the second correction condition. The correction condition for correcting the correction target image can be calculated so that each of the first predetermined values becomes a second predetermined value of the gradation value corresponding to the shadow in the correction target image. As a result, even when the correction target image is an image taken under illumination with a different light source, a correction condition that can correct the color balance of the correction target image properly can be obtained. Correction conditions that can appropriately correct the density and color balance of an image with a very small range can be obtained.
[0014]
Further, in the first aspect of the invention, the first correction condition and the second correction condition may be an arithmetic expression of the correction calculation, but the first correction condition and the second correction condition are correction targets. In the case where the conversion condition is for correcting the correction target image by converting the image, the correction means, for example, as described in claim 4, the conversion condition corresponding to the first correction condition and the second A conversion condition corresponding to the weighted average of the conversion conditions corresponding to the correction condition is obtained, and the correction target image is converted according to the obtained conversion condition, so that the correction corresponding to the weighted average can be performed.
[0015]
Further, in the invention according to claim 4, the correction means can be configured to include a three-dimensional lookup table. In this case, the correction means, for example, as described in claim 5, sets the obtained conversion condition to the 3 Correction can be performed by inputting correction target image data after being set in the dimension lookup table. Thereby, the correction corresponding to the weighted average of the correction under the first correction condition and the correction under the second correction condition can be realized by a simple process.
[0016]
According to a sixth aspect of the present invention, there is provided an image correction method for recognizing a main part region of a correction target image and setting a first correction condition for correcting the correction target image based on a feature amount of the recognized main part region. A second correction condition for calculating and correcting the correction target image so that a difference between a gradation value corresponding to highlight and a gradation value corresponding to shadow in the correction target image becomes a predetermined value And calculating a weight for the first correction condition and the second correction condition based on the dynamic range of the correction target image, and calculating the calculation for the correction target image based on the set weight. Since the correction corresponding to the weighted average of the correction under the first correction condition and the correction under the calculated second correction condition is performed, an image with a very small dynamic range is provided as in the invention of claim 1. It is possible that concentration and the like are also appropriately corrected.
[0017]
According to a seventh aspect of the present invention, there is provided a program for recognizing a main area of a correction target image and correcting the correction target image based on a feature amount of the recognized main area. A second computing unit for correcting the correction target image so that a difference between a gradation value corresponding to highlight and a gradation value corresponding to shadow in the correction target image becomes a predetermined value. Second calculation means for calculating the correction condition, setting means for setting weights for the first correction condition and the second correction condition based on the dynamic range of the correction target image, and weights set by the setting means Based on the above, the correction target image is corrected under the first correction condition calculated by the first calculation means, and under the second correction condition calculated by the second calculation means. To function as a correction means for correcting corresponding to the weighted average of the positive.
[0018]
According to a seventh aspect of the invention, a program (for example, a computer incorporated in a photographing apparatus such as a DSC or other equipment) may be used as the first calculation means, the second calculation means, and the setting means. Therefore, when the computer executes the program according to the seventh aspect of the present invention, the computer functions as the image correction apparatus according to the first aspect. Similar to the invention, it is possible to appropriately correct the density of an image having a very small dynamic range.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a color printer 10 to which the present invention is applied. The color printer 10 is configured to record an image on a long color thermosensitive recording paper 12 provided with each thermosensitive coloring layer. The color thermosensitive recording paper 12 is a recording paper roll 13 wound in a roll shape, and the apparatus main body 11. Set inside.
[0020]
The color thermal recording paper 12 is sandwiched between a pair of paper feed rollers 14, and the driving force of the motor 16 is transmitted to rotate the paper feed roller 14 to feed the paper downstream. A thermal head 18 to which a heating element array 19 in which a plurality of heating elements are arranged in a line is attached is disposed downstream of the paper feed roller 14. The thermal head 18 is rotatable about a rotary shaft 20, and a recording position (a position indicated by a broken line) where the heating element array 19 is pressure-bonded to the color thermal recording paper 12 by transmitting a driving force of a motor (not shown). ) Or a retreat position (a position indicated by a solid line) away from the color thermal recording paper 12.
[0021]
Further, a platen roller 22 is disposed at a portion facing the heating element array 19. The platen roller 22 rotates according to the conveyance of the color thermal recording paper 12 and supports the color thermal recording paper 12 to improve the stability of the contact state between the color thermal recording paper 12 and the heating element array 19. When recording an image on the color thermal recording paper 12, the thermal head 18 causes each heating element of the heating element array 19 to generate heat at a predetermined temperature in accordance with print data (described later), and each thermal coloring layer of the color thermal recording paper 12 is formed. Selectively develop color.
[0022]
A pair of transport rollers 26 for transporting the color thermal recording paper 12 at the time of image recording is disposed on the downstream side of the thermal head 18. The transport roller 26 is rotated by the driving force of the motor 16 being transmitted. Then, the color thermal recording paper 12 is transported by the paper feed roller 14 and the transport roller 26 in the paper feeding direction pulled out from the recording paper roll 13 or in the rewinding direction rewound on the recording paper roll 13.
[0023]
A fixing device 28 is disposed between the thermal head 18 and the transport roller 26. The fixing device 28 includes a yellow ultraviolet lamp 29, a magenta ultraviolet lamp 30, and a reflector 31 for reflecting the ultraviolet rays emitted from the ultraviolet lamps 29, 30 toward the color thermal recording paper 12. The yellow heat-sensitive color developing layer and the magenta heat-sensitive color developing layer on the color heat-sensitive recording paper 12 are fixed by the ultraviolet light emitted from the ultraviolet lamps 29 and 30 so as not to be colored even if heated again.
[0024]
A print discharge port 34 is formed on the downstream side of the conveyance roller 26, and a cutting device 32 for cutting the color thermal recording paper 12 is disposed between the print discharge port 34 and the conveyance roller 26. The color thermal recording paper 12 cut by the cutting device 32 is discharged to the outside of the main body 11 through the print discharge port 34.
[0025]
Next, the configuration of the control system of the color printer 10 will be described with reference to FIG. The color printer 10 includes a CPU 60 and an I / O circuit 62 that control the operation of the entire apparatus, and a RAM 86 and a ROM 88 are connected to the CPU 60. In the storage area of the RAM 86, a work area for temporarily storing data when the CPU 60 performs various processes is formed. The ROM 88 stores an image correction program for performing an image correction process described later. This image correction program corresponds to the program described in claim 7.
[0026]
Further, a digital still camera (DSC) 64 (recording medium) and a computer 66 are connected to the I / O circuit 62, and image data (for example, the color of each pixel is 8 bits each for R, G, B) from the DSC 64 or the computer 66. Is input). A frame memory 68 is connected to the I / O circuit 62, and image data input to the I / O circuit 62 is stored in the frame memory 68. The I / O circuit 62 is provided with a video output terminal. The display 70 connected to the I / O circuit 62 via the video output terminal has image data output via the video output terminal. An image represented by (NTSC composite signal) is displayed.
[0027]
The frame memory 68 is connected to a memory controller 72 and a print data forming unit 74. The image data stored in the frame memory 68 is corrected by the image correction processing performed by the CPU 60 and then read from the frame memory 68 by the CPU 60 controlling the memory controller 72, and the print data forming unit 74. Is output. The print data forming unit 74 generates print data for driving the heating element array 19 based on the image data input from the frame memory 68. The print data generated by the print data forming unit 74 is supplied to the head driver 76. The head driver 76 causes each heating element of the heating element array 19 to generate heat according to the input print data. As a result, an image is recorded on each thermosensitive coloring layer of the color thermosensitive recording paper 12 by each heater element of the heater element array 19.
[0028]
Next, as an operation of the present embodiment, image data representing an image to be recorded on the color thermal recording paper 12 is input from the DSC 64 or the computer 66 and stored in the frame memory 68, and an image correction program stored in the ROM 88 is stored. The image correction process realized by being executed by the CPU 60 will be described with reference to the flowchart of FIG.
[0029]
In step 100, the processing target image data stored in the frame memory 68 is read, and the read processing target image data is thinned out in units of pixels, thereby reducing the image represented by the original image data (for example, 128 pixels). Reduced image data representing thumbnail images of about 92 pixels × 160 pixels × 120 pixels) is generated. In step 102, an R, G, B histogram is generated from the reduced image data generated in step 100, and the reduced image data is converted into YCbCr color system data according to a predetermined arithmetic expression, and obtained by conversion. A histogram of Y, Cb, and Cr is generated from Y, Cb, and Cr data Y, Cb, and Cr.
[0030]
In the next steps 104 to 124, conversion data corresponding to the first correction condition according to the present invention is obtained. That is, in step 104, the color balance correction amount is set based on the YCbCr color system data Y, Cb, Cr obtained from the reduced image data. Specifically, first, a gray detection zone extending along the Y axis is set on the YCbCr color space as shown in FIG. 4 as an example. In the example of FIG. 4, two types of gray detection zones (gray detection zone (inner side) and gray detection zone (outer side)) each having an ellipsoidal shape and different radii in the vicinity of the intermediate luminance are shown as gray detection zones. ing.
[0031]
Next, based on the data Y, Cb, and Cr obtained from the reduced image data, among the pixels on the reduced image data, pixels whose positions on the YCbCr color coordinates are located in the gray detection zones are determined. Extracting and obtaining the color balance GrayIn_Cb, GrayIn_Cr of the pixel located in the gray detection zone (inner side), located outside the gray detection zone (inner side) and located in the gray detection zone (outer side) The color balances GrayOut_Cb and GrayOut_Cr of the current pixel are obtained. Then, the gray balance Gray_Cb, Gray_Cr of the image to be processed is obtained according to the following equation (1).
Gray_Cb = (GrayIn_Cb + GrayOut_Cb) / 2
Gray_Cr = (GrayIn_Cr + GrayOut_Cr) / 2 (1)
As the gray balance of the image to be processed, instead of obtaining the average color balance of the pixels located in each gray detection zone as described above, a weight is assigned to each gray detection zone. The weighted average of the color balance of the pixels located in each gray detection zone may be obtained.
[0032]
Further, as the color balance correction amounts Cb_offset and Cr_offset, the gray balance Gray_Cb and Gray_Cr of the image to be processed may be used as they are, and the gray balance Gray_Cb and Gray_Cr may be set to 0, but the image data to be processed Is image data input from the DSC 64 (for example, an image file in EXIF format or other format), the image data includes shooting information representing shooting conditions at the time of shooting. Values obtained by multiplying the gray balance Gray_Cb, Gray_Cr of the image to be processed by the coefficient determined based on the photographing brightness BV (Brightness Value) are used as the color balance correction amounts Cb_offset, Cr_offset (see the following equation (2)). You may do it.
Cb_offset = Gray_Cb × f (BV)
Cr_offset = Gray_Cr × f (BV) (2)
[0033]
In the above equation, the function f (BV) is an intensity coefficient whose value changes according to the photographing luminance BV. For example, when BV <0 (night scene or fireworks scene), f (BV) = 1.0. , 0 ≦ BV <1 (scene under dim tungsten light source, fluorescent lamp, mixed light source), f (BV) = 0.1 × BV + 1, 1 ≦ BV <3 (normal indoor scene) Is f (BV) = 1.1, 3 ≦ BV <6 (daytime outdoor scene or backlit scene), f (BV) = − 2 × BV / 15 + 1.5, 6 ≦ BV (daytime in fine weather) In the case of an outdoor or seamount scene), f (BV) = 0.7 can be set.
[0034]
In addition, the shooting information includes information indicating whether or not shooting was performed using a strobe, and the intensity coefficient is changed according to whether or not the strobe was used during shooting. It may be. For example, in the above example, the intensity coefficient is set to a value larger than 1 when the shooting brightness BV is a value corresponding to a slightly dark scene, but when the image is shot using a strobe, the gray balance of the image is appropriate. Since there are relatively many cases, the value of the intensity coefficient may be reduced (for example, a value of about 0.8).
[0035]
In step 106, a threshold for dividing the reduced image into a plurality of areas is set. That is, first, the luminance difference between the specific pixel P0 and the eight neighboring pixels P1 to P8 (see FIG. 5) around the specific pixel P0 is calculated for the specific pixel P0 among the pixels of the reduced image. An average value (average luminance difference) is calculated. Further, the color difference between the specific pixel P0 and the eight neighboring pixels P1 to P8 is calculated, and the average value (average color difference) is also calculated. Then, the above calculation is performed for each pixel of the reduced image, and the average value of the average luminance difference for each pixel (in-screen average luminance difference Ymean) is obtained.
[0036]
In the present embodiment, dividing the reduced image into a plurality of regions is whether or not each pixel of the reduced image is a pixel corresponding to the boundary of the region based on whether or not the average luminance difference and the average color difference are equal to or greater than a threshold value. Do by judging. For this reason, the number of divided areas when the reduced image is divided changes according to the threshold value. For example, in FIG. 6, a value obtained by multiplying the average luminance difference in the screen Ymean by a coefficient is used as a threshold value for the average luminance difference, and the number of divided regions is obtained by performing region division on a large number of natural images (reduced images). The relationship between the coefficient value and the number of divided areas (maximum area number, average area number, and minimum area number) when this is repeated while changing the coefficient value is shown.
[0037]
According to the inventor of the present application, it is found that if the reduced image is divided into about 20% of the total number of pixels of the reduced image, the area division reflecting the characteristics of the original image to be processed is possible. doing. Therefore, the threshold value of the average luminance difference can be obtained by determining a coefficient using the relationship shown in FIG. 6 based on the total number of pixels of the reduced image and multiplying the determined coefficient by the average luminance difference in the screen Ymean.
[0038]
For example, when the reduced image is an image of 160 pixels × 120 pixels, the total number of pixels is 19200, and the number of appropriate divided regions is 3840, which is 20% of the total number of pixels. In the relationship between the coefficient values shown in FIG. 5 and the number of divided areas (average area number), the coefficient value when the number of divided areas is 3840 is 0.5. A value obtained by multiplying 0.5 can be used as a threshold value of the average luminance difference. The average color difference threshold value may be fixed in advance or may be changed according to the average color difference average value (in-screen average color difference) for each pixel.
[0039]
In the next step 108, the reduced image is divided into a plurality of regions using the threshold value determined in step 106. Specifically, for example, the average luminance difference and the average color difference of adjacent pixels on the reduced image are compared, and if the difference in average luminance difference is equal to or smaller than the threshold and the difference in average color difference is equal to or smaller than the threshold, the adjacent pixels Repeating scanning the reduced images in order is performed by setting the pixels in the same region. As a result, the reduced image is divided into a plurality of regions each having substantially uniform luminance and hue within the region. When the region division is completed, feature amounts such as average density and saturation are calculated for each region.
[0040]
In step 110, a region estimated not to be a main region is removed from a plurality of regions obtained by the division. That is, first, the saturation of each region is compared with a predetermined value, and regions where the saturation is greater than or equal to the predetermined value are removed. Subsequently, the first limit concentration DFmin is obtained according to the following equation (3), and the second limit concentration DFmax is obtained according to the equation (4).
DFmin = (Dmin + Dstd) / 2 (3)
DFmax = (Dmax + Dstd) / 2 (4)
However, Dmin is the lowest density of the reduced image, Dmax is the highest density of the reduced image, and Dstd is the reference density (for example, Dstd = 0.75). Next, out of the unremoved regions, a region having an average density equal to or higher than the first limit density DFmin or equal to or lower than the second limit density DFmax is extracted as a removal candidate area.
[0041]
Next, for each area extracted as a removal candidate area, the density feria moment dfmnt is calculated according to the following equation (5).
dfmnt = | dav−Dstd | × s (5)
Where dav is the average density of the removal candidate region, and s is the area of the removal candidate region. Then, the density feria / moment dfmnt values of the respective removal candidate areas are compared, and a predetermined number of removal candidate areas are removed in descending order of the density feria / moment dfmnt value. Therefore, the processing in step 110 removes a high-saturation region and removes a region with a relatively large area at low luminance (high density) or high luminance (low concentration).
[0042]
In step 112, the average density Dav of the entire region remaining without being removed through the processing of step 110 is calculated. In step 114, a skin color area is extracted from the remaining areas as the main area, and the average density Ds and area ratio Rs of the extracted main area are calculated. In addition to the above-mentioned condition that “hue is flesh-colored”, for example, conditions such as “the area is within a predetermined value”, “the center position is separated from the four corners of the reduced image by a predetermined value”, “not an elongated shape”, and the like are added. Thus, the main area may be extracted.
[0043]
In the next step 116, the correction target density Dtarg is calculated. This calculation is performed by first converting the area ratio Rs of the main area into the correction coefficient coef (Rs) according to the conversion conditions as shown in FIG. 7 as an example, and then the average density Dav, the average density Ds of the main area, and the correction coefficient. This can be realized by substituting coef (Rs) into the following equation (6).
Dtarget = [[3-2 × coef (Rs)] × Dav + 2 × coef (Rs) × Ds] / 3 (6)
In step 118, a density correction curve for converting the correction target density Dtarg obtained in step 116 into a fixed reference density Dstd is set. Thereby, a density correction curve as shown in FIG. 8 is obtained as an example. Instead of the reference density Dstd, a target density Dtarget defined by the following equation (7) may be used.
Dtarget = Dstd−C1 × coef (Rs) (7)
However, C1 is a constant whose value is 0 or 1 depending on whether or not the image to be processed is an image obtained by photographing a backlight scene.
[0044]
In step 120, a brightness correction curve is generated based on the density correction curve set in step 118. That is, first, in the histogram of Y generated in step 102, the luminance Ymax corresponding to the point (highlight point) at which the cumulative number of pixels from the maximum value of the luminance Y becomes 0.4% of the total number of pixels is obtained. A luminance Ymin corresponding to a point (shadow point) at which the cumulative number of pixels from the minimum value of Y becomes 0.4% of the total number of pixels is obtained.
[0045]
For example, as shown in FIG. 9, when the luminance input value Yin is the luminance Ymax corresponding to the highlight point, the luminance input value Yin is converted to the maximum luminance (for example, 255 if the luminance output value Yout is 8 bits). When the luminance Ymin corresponding to the shadow point is converted to the minimum luminance (for example, the luminance output value Yout is 0), and when the luminance input value Yin is an intermediate gradation value corresponding to between the luminance Ymax and the luminance Ymin, a step is performed. The conversion characteristic of the luminance correction curve is set so that it is converted into the luminance output value with the conversion characteristic according to the density correction curve set in 118, and the luminance correction curve is generated so that the luminance is converted with the set conversion characteristic. To do.
[0046]
As a result, the luminance input value Yin is subjected to density correction for converting the correction target density Dtarg obtained by extracting the main region into the reference density Dstd (at this time, the contrast correction is also performed simultaneously). As a result, a luminance correction curve to be output as is obtained. In step 122, the saturation correction amount is calculated based on the reduced image data. Although the description of the calculation of the saturation correction amount is omitted, for example, the technique described in JP-A-2001-218078 can be applied.
[0047]
In step 124, the color balance correction amount set in step 104, the luminance correction curve set by the processing in steps 106 to 120, and the saturation correction amount calculated in step 122 are shown in FIG. In this way, the input R, G, B data is converted into YCbCr color system data, and the data Y corresponding to the density correction (and contrast correction) corresponding to the luminance correction curve is performed on the data Y. Is converted to data Cb ′ and Cr ′ corresponding to the color balance correction amount corresponding to the color balance correction amount for the data Cb and Cr, and further the data Cb ′ and Cr ′. The data is converted into data Cb ″, Cr ″ corresponding to the saturation correction corresponding to the saturation correction amount, and the converted data of Y ′, Cb ″, Cr ″ is R ′, G ′, B ′. To convert the data, sets the conversion data for realizing the conversion of one (corresponding to the first correction condition according to the present invention).
[0048]
The above conversion is performed using a three-dimensional lookup table (3D-LUT). In this embodiment, the amount of conversion data (data defining output data stored in association with input data) is determined. In order to reduce the conversion data, thinning of the conversion data is performed, and each of the 8-bit data of R, G, B each divides the numerical range (0 to 255) that can be expressed by 8-bit data into 8 increments. Only conversion data in the case of a value corresponding to the division position is stored, and if there is no conversion data corresponding to R, G, B data input to the 3D-LUT, the stored conversion is stored. Output data is obtained from the data by interpolation.
[0049]
For this reason, the setting of the conversion data in step 124 is represented by 8-bit data of R, G, and B distributed on the RGB color space defined by the grid points (R, G, and B coordinate axes orthogonal to each other). By dividing the possible color reproduction range at a position corresponding to the boundary when 256 gradations that can be expressed by 8-bit data are divided every 8 gradations, the color reproduction range is divided into a large number of cubes in a lattice shape. Only the R, G, B data corresponding to the R, G, B data corresponding to the vertices of the individual rectangular areas when divided into areas (R, G, B → R ′, G ′, B ′). Conversion) and associating the converted data with the data before conversion.
[0050]
As described above, it is possible to obtain conversion data that can appropriately correct the density, color balance, saturation, and the like of the majority of images excluding images with a very small dynamic range. Steps 104 to 124 described above correspond to the first calculation means according to the present invention (specifically, the first calculation means according to claim 2).
[0051]
In the next steps 126 to 130, processing for obtaining conversion data corresponding to the second correction condition according to the present invention is performed. That is, in step 126, the highlight point from the histogram of R, G, B generated in step 102 (the point at which the cumulative number of pixels from the minimum of density R or density G or density B becomes 0.4% of the total number of pixels). Concentrations Rmax, Gmax, and Bmax corresponding to are extracted. In step 128, the brightness Y (Y = 0.299 × R + 0.587 × G + 0.114 × B) generated in step 102 is used to calculate the shadow point (the cumulative number of pixels from the minimum value of luminance Y is the total number of pixels). A gradation value Ymin corresponding to 0.4% is obtained.
[0052]
In step 130, for each of the R, G, and B color components, as shown in FIG. 11 as an example, the density value Rmax or density value Gmax or density value Bmax corresponding to the highlight point is 255, and the level corresponding to the shadow point. A conversion condition for converting the tone value Ymin to 0 and linearly converting the intermediate gradation is determined, and the input R, G, B data is converted to R ′, G ′, B ′ according to the conversion condition defined above. Conversion data (corresponding to the second correction condition according to the present invention) for realizing conversion to data by the 3D-LUT is set. Note that the above conversion condition is represented by an arithmetic expression as follows.
R ′ = 255 × (R−Ymin) / (Rmax−Ymin)
G ′ = 255 × (G−Ymin) / (Gmax−Ymin)
B ′ = 255 × (B−Ymin) / (Bmax−Ymin)
Also, with respect to the setting of the conversion data in step 130, only the R, G, B data corresponding to the grid points are converted based on the determined conversion conditions (R, G, B → R ′, G ′, B ′). Conversion), and the converted data is associated with the data before conversion.
[0053]
By the above processing, conversion data that realizes a conversion condition (conversion condition that maximizes the dynamic range) that maximizes the difference between the density value corresponding to the highlight point and the density value corresponding to the shadow point (255) is obtained. Therefore, by converting the image data according to the conversion data, it is possible to appropriately correct the density of the image having a very small dynamic range. The conversion condition realized by the conversion data is a conversion condition for converting the density values Rmax, Gmax, and Bmax corresponding to the highlight points to the same value (= 255). Even if the highlight point in the image to be processed has a color peculiar to a different light source, for example, because the image is taken under illumination with a different light source, the conversion using the conversion data is performed. Thus, the highlight point is corrected to be white, and the different light source correction is also performed at the same time. Steps 126 to 130 described above correspond to the second calculation means according to the present invention (specifically, the second calculation means according to claim 3).
[0054]
In step 132, the maximum value dmax and the minimum value dmin of Y luminance in the reduced image are obtained, and the dynamic range d of the image is calculated according to the following equation (8).
d = dmax−dmin (8)
In step 134, the weight coefficient c (d) for the conversion data corresponding to the first correction condition is obtained based on the dynamic range d obtained in step 132. As the weighting coefficient c (d), for example, as shown in FIG.
When d ≦ d1: c (d) = 0
When d1 <d <d2: c (d) = (d−d1) / (d2−d1)
When d ≧ d2: c (d) = 1
A function having the following characteristics can be used. Then, based on the obtained weight coefficient c (d), conversion data corresponding to a weighted average of conversion data corresponding to the first correction condition and conversion data corresponding to the second correction condition is generated.
[0055]
Note that the generation of the conversion data can also be realized by performing the calculation of the following equation (9) for each lattice point in setting the conversion data.
f (r, g, b) = c (d) * p (r, g, b) + (1-c (d)) * q (r, g, b) (9)
However, p (r, g, b) is conversion data corresponding to a specific lattice point among the conversion data corresponding to the first correction condition, and q (r, g, b) is equivalent to the second correction condition. Among the conversion data, conversion data corresponding to a specific grid point, f (r, g, b) is conversion data corresponding to a weighted average corresponding to the specific grid point.
[0056]
As described above, the conversion data generated in step 134 is obtained by converting the conversion data corresponding to the first correction condition and the conversion data corresponding to the second correction condition into weight coefficients c (d) corresponding to the dynamic range of the image. Conversion data representing conversion conditions corresponding to weighting and averaging using the conversion data, and by converting the image data to be processed using the conversion data, the dynamic range of the image to be processed is considered to be large or small. Therefore, it is possible to appropriately correct the density of the image to be processed.
[0057]
Note that the processing for obtaining the weighting coefficient c (d) in Step 132 and Step 134 described above corresponds to the setting means according to the present invention, and the processing for obtaining conversion data corresponding to the weighted average in Step 134 includes In addition to the process of actually converting the image data to be processed in accordance with the conversion data, this corresponds to the correction means according to the present invention (specifically, the correction means described in claims 4 and 5).
[0058]
When the image correction process described above is completed, the CPU 60 sets the conversion data obtained by the image correction process in a conversion table that functions as a 3D-LUT. Then, the image data to be processed stored in the frame memory 68 is sequentially extracted in units of individual pixel data, and sequentially converted based on a conversion table in which conversion data is set.
[0059]
In this conversion, first, conversion data that associates the extracted pixel data (8-bit data for each of R, G, and B) with output data is registered in the conversion table (that is, whether the data corresponds to the grid point position). If it is registered in the conversion table, output data (R, G, B, each 8 bits data) registered in the conversion table in association with the pixel data is output. If not registered in the conversion table, a plurality of outputs registered in the conversion table in association with data of a plurality of grid point positions located around the coordinate position of the pixel data in the RGB color space Based on the data, output data corresponding to the pixel data is obtained by interpolation calculation. By the above conversion, the density or the like of the image to be processed is corrected appropriately regardless of the dynamic range of the image to be processed.
[0060]
The image data to be processed that has undergone the conversion by the 3D-LUT is temporarily stored in the frame memory 68, then input to the print data forming unit 74, and converted into print data by the print data forming unit 74. This print data is supplied to the head driver 76, and the head driver 76 generates heat in each heat generating element of the heat generating element array 19 according to the print data, whereby an image is recorded on each heat sensitive coloring layer of the color thermosensitive recording paper 12. . The color thermal recording paper 12 on which the image is recorded is fixed by the fixing device 28, then cut by the cutting device 32 in units of image defects, and discharged to the outside of the apparatus main body 11 through the print discharge port 34. Will be.
[0061]
In the above description, conversion data for realizing correction corresponding to the weighted average of correction under the first correction condition according to the present invention and correction under the second correction condition according to the present invention is obtained, and the obtained conversion data is obtained. As an example, a mode corresponding to the weighted average is applied to the processing target image data by converting the processing target image data by setting the 3D-LUT to the processing target. However, the present invention is not limited to this. Rather than being calculated, an arithmetic expression for performing correction corresponding to the weighted average is obtained, and correction corresponding to the weighted average is performed by performing correction calculation on the image data to be processed according to the arithmetic expression. You may do it.
[0062]
In the above description, the example in which the present invention is applied to the correction of image data in the color printer 10 configured to record an image on the color thermal recording paper 12 has been described. However, the present invention is widely applicable to the correction of digital images. For example, it can be converted into data by reading an image recorded on photographic film and applied to correction when creating a photographic print by exposure recording on photographic paper, or by imaging a subject with a digital still camera (DSC) or the like. It is also possible to apply the correction to the obtained image data (for example, image correction when performing photo processing requested via a network such as the Internet).
[0063]
The DSC stores an image correction program for realizing the image correction process described above and a conversion program for correcting the image data by converting the image data based on the conversion data generated by the image correction process. By doing so, the image correction according to the present invention may be performed by the DSC. In this aspect, the DSC functions as the image correction apparatus according to the present invention.
[0064]
Further, the image correction software and the conversion program described above may be added to image processing software for execution on a personal computer (PC) or the like so that the image correction according to the present invention is performed on the PC. In this aspect, the PC functions as the image correction apparatus according to the present invention. The image processing software also corresponds to the program according to claim 7.
[0065]
【The invention's effect】
As described above, the present invention corresponds to the first correction condition that is recognized based on the feature amount of the recognized main part area, and the highlight in the correction target object image. A weight for the second correction condition calculated so that the difference between the gradation value and the gradation value corresponding to the shadow becomes a predetermined value is set based on the dynamic range of the correction target image. Since the correction corresponding to the weighted average of the correction under the first correction condition and the correction under the second correction condition is performed, it is possible to appropriately correct the density of an image having a very small dynamic range. , Has an excellent effect.
[Brief description of the drawings]
FIG. 1 is a schematic configuration diagram of a color printer according to an embodiment.
FIG. 2 is a block diagram illustrating a schematic configuration of a control system of a color printer.
FIG. 3 is a flowchart showing the contents of image correction processing.
FIG. 4 is a conceptual diagram illustrating an example of a gray detection zone.
FIG. 5 is a conceptual diagram illustrating a specific pixel and eight neighboring pixels around the specific pixel.
FIG. 6 is a diagram showing an example of a relationship between a coefficient value multiplied by an in-screen average luminance difference Ymean and the number of divided regions (maximum number of regions, average number of regions, and minimum number of regions).
FIG. 7 is a diagram showing an example of a relationship between an area ratio Rs of a main region and a correction coefficient coef (Rs).
FIG. 8 is a diagram showing an example of a density correction curve.
FIG. 9 is a diagram showing an example of a luminance correction curve.
FIG. 10 is a conceptual diagram showing an outline of conversion according to a first conversion condition.
FIG. 11 is a diagram showing an example of conversion characteristics of a second conversion condition.
FIG. 12 is a diagram showing an example of a relationship between a dynamic range d and a weighting factor c (d).
[Explanation of symbols]
10 Color printer
12 color thermal recording paper
18 Thermal head
28 Fixing device
32 Cutting device
60 CPU
68 frame memory

Claims (7)

First calculating means for recognizing a main part region of the correction target image and calculating a first correction condition for correcting the correction target image based on a feature amount of the recognized main part region;
A second calculation for calculating a second correction condition for correcting the correction target image so that a difference between a gradation value corresponding to highlight and a gradation value corresponding to shadow in the correction target image becomes a predetermined value. Computing means;
Setting means for setting weights for the first correction condition and the second correction condition based on the dynamic range of the correction target image;
Based on the weight set by the setting unit, the correction target image is corrected under the first correction condition calculated by the first calculation unit and the second correction calculated by the second calculation unit. Correction means for performing correction corresponding to a weighted average of corrections under conditions;
An image correction apparatus including: The first calculation means sets a color balance correction condition based on a distribution on a color coordinate of a pixel group having a saturation equal to or less than a predetermined value in the correction target image, and also recognizes a feature amount of a recognized main region. A correction condition for density is set so that the correction target density obtained from the above is converted to a predetermined target density, and the color balance and density of the correction target image are corrected under the set correction condition as the first correction condition. The image correction apparatus according to claim 1, wherein the condition is calculated. The second calculation means has, as the second correction condition, gradation values corresponding to highlights for the respective color components in the correction target image, respectively, become first predetermined values, and corresponds to shadows in the correction target image. The image correction apparatus according to claim 1, wherein a correction condition for correcting the correction target image is calculated so that a gradation value to be adjusted becomes a second predetermined value. The first correction condition and the second correction condition are conversion conditions for correcting the image to be corrected by converting the correction target image, and the correction means includes a conversion condition corresponding to the first correction condition and a second correction condition. A conversion condition corresponding to a weighted average of conversion conditions corresponding to the correction condition is obtained, and correction corresponding to the weighted average is performed by converting a correction target image according to the obtained conversion condition. The image correction apparatus according to 1. The correction unit includes a three-dimensional lookup table, and performs the correction by inputting data of a correction target image after setting the obtained conversion condition in the three-dimensional lookup table. The image correction apparatus according to claim 4. Recognizing the main part region of the correction target image, calculating a first correction condition for correcting the correction target image based on the feature amount of the recognized main part region,
In addition, a second correction condition for correcting the correction target image is calculated so that a difference between the gradation value corresponding to the highlight in the correction target image and the gradation value corresponding to the shadow becomes a predetermined value. With
Setting weights for the first correction condition and the second correction condition based on the dynamic range of the correction target image;
An image correction method for performing correction corresponding to a weighted average of correction under the calculated first correction condition and correction under the calculated second correction condition on the correction target image based on a set weight. Computer
A first calculation means for recognizing a main part region of the correction target image and calculating a first correction condition for correcting the correction target image based on a feature amount of the recognized main part region;
A second calculation for calculating a second correction condition for correcting the correction target image so that a difference between a gradation value corresponding to highlight and a gradation value corresponding to shadow in the correction target image becomes a predetermined value. Computing means,
Setting means for setting weights for the first correction condition and the second correction condition based on a dynamic range of the correction target image;
Based on the weight set by the setting unit, the correction target image is corrected under the first correction condition calculated by the first calculation unit and the second correction calculated by the second calculation unit. A program for functioning as correction means for performing correction corresponding to a weighted average of correction under conditions.

Images