JP4881325B2 - Profiling digital image input devices - Google Patents

Description

  The present invention relates to a profiling technique for obtaining color characteristics (color profile) of a digital image input device used to obtain data based on digital data representing a color sample (color chart) image. According to the present invention, the color profile of the input device can be obtained regardless of the relationship between the color sample partial exposure degree and the non-color sample partial exposure degree in the image.

  Examples of digital image input devices include digital cameras and scanners. This type of apparatus captures reflected light from a subject and converts it into digital image data. Image data obtained by the conversion represents an image of a subject and is composed of a plurality of data units. Since each data unit corresponds to each constituent part of the image, on the data, the image is expressed in units of the constituent part, that is, the pixel, and therefore expressed as a two-dimensional array of pixels (coordinate values: x, y). become. In addition, the color representation in the image data is performed for each data unit, that is, in units of pixels. Usually, the pixel is determined by the amount of the primary color component in the pixel, that is, the intensity of the red (R), green (G) and blue (B) components. The color is expressed or described. For example, the color of the pixel located at x = 0, y = 0 is expressed by color intensity data such as R = 200, G = 134, and B = 100. This is an example in which the intensity of each primary color component is represented by 8 bits. With 8 bits, 256 primary color component intensities from 0 to 255 can be expressed. Usually, the maximum strength is represented by 255 and the minimum strength (black) is represented by 0.

  Since the color characteristics of the input device appear in the image data obtained from the digital image input device, in the present application, such image data will be referred to as device-dependent data. That is, even when the same image is taken under the same conditions, the value of the obtained image data is usually different if the input device used for it is different. For example, when the same subject is photographed under the same conditions, a pixel with R = 200 in an image taken with one digital camera may be R = 202 with an image taken with another digital camera. . More specifically, when the same apple is photographed under the same conditions, an image of the red part of the apple is recorded as an image of R = 200 in the former, whereas an image of the same part is recorded as R = 200 in the latter. That is, it is recorded as 202 images. The device-dependent data generated by the input device is usually data representing R, G, and B component values in each pixel and is called RGB data.

  The discrepancy in device-dependent data between input devices is due to slight differences between the various imaging members incorporated in those input devices, and the device-dependent data obtained by these input devices Therefore, various problems are caused when an image is output by the same digital image output device, for example, a color ink jet printer, a CRT (cathode ray tube) monitor, an LCD (liquid crystal) monitor, or the like. For example, if the image taken by the two digital cameras shown in the previous example is printed by the same color inkjet printer, the printed result looks different even though the apples taken are the same.

  To complicate matters, there are similar differences between digital image output devices as well as differences between digital image input devices. For example, a red square image (for example, an image with all pixels R = 200, G = 0, and B = 0) provided as data from a common input device is displayed on a commercial CRT monitor and a customer CRT monitor. Are displayed simultaneously. In that case, although the display is based on the same digital data given from the same input device, the displayed images are not always the same, and there is usually a slight difference in red for each monitor.

  Further, when an image is printed by a plurality of printers based on the same image data, the same difference is usually generated. However, it should be noted that the printer prints an image based on the CMYK data. The CMYK data is digital image data that expresses or describes the color of each pixel based on the intensity of each derived color component, that is, each color component of cyan (C), magenta (M), yellow (Y), and black (K) component. . This CMYK data also has device dependence, like RGB data obtained from a digital image input device and RGB data used for monitor display.

  Such a color difference due to a difference in apparatus can be dealt with by a color profile. The color profile is information describing the color characteristics of each device. For example, the input device converts (maps) device-dependent data generated by itself into device-independent data based on its own color profile. Similarly, the output device converts the given device-independent data into device-dependent data for itself based on its own color profile, and performs printing based on the obtained device-dependent data, thereby expressing the device-independent data. Reproduce the color to be printed on the printed material. The device-independent data is data that describes or expresses the color of each pixel in a universal device-independent color space. The device-independent color space is a space in which each color is represented by a unique value (a value determined under a specific device state and lighting condition) corresponding to the color. In the industry, CIEXYZ, CIELAB, CIEXY, CIECH (CIE: International Lighting Commission). In this application, device-independent data is also referred to as coordinate values in the device-independent color space, device-independent data conforming to CIEXYZ color space is referred to as XYZ data, CIEXYZ, or XYZ, and device-independent according to CIELAB color space. The data may be referred to as LAB data, CIELAB, or LAB.

  In principle, the user can cause the digital image output device to output an accurate mapping of the subject image captured by any digital image input device by using the color profile. For example, as shown in FIG. 1, first, an image of a subject 101 is captured by a digital image input device 102, and device-dependent RGB data 103 representing the result is generated. Next, the data 103 is converted into device-independent XYZ data 105 based on the color profile 104 of the input device 102. Subsequently, the data 105 is converted into device-dependent CMYK data 107 based on the color profile 106 of the digital image output device 108 to be used. The output device 108 generates an accurate map 109 of the subject 101 based on the data 107.

  In obtaining a color profile, first, a color sample is first photographed under reference illumination conditions, and the image, that is, the color sample image is usually acquired as a test image. For example, when obtaining a color profile of a digital camera, a test image is generated by taking a photograph of a cocoon decorated with fruits, for example, as a color sample and photographing it with its background. A color sample is not limited to a fruit basket image, and generally includes a plurality of patch-like portions (color patches) having different colors, and the color when each color patch is observed under a predetermined reference illumination condition. Any tangible object can be used as long as it has a predetermined color. For example, a D50 condition known in the industry is used as the reference illumination condition, and a device-independent color space such as a CIELAB color space is used as a space for describing the color of the color patch under the reference illumination condition. In any case, the device-dependent data (for example, the color obtained by photographing) obtained by the test image photographing is used as the color of each part of the color sample that is known in advance (defined by the coordinate value in the device-independent color space). By comparing, the color profile of the apparatus used for acquiring the test image can be obtained.

US Patent Application Publication No. 2002/0167528 (A1) US Patent Application Publication No. 2004/018381 (A1) US Pat. No. 6,108,442 US Pat. No. 6,459,436 US 2003/0053085 (A1) US 2002/0051159 (A1) US Pat. No. 5,818,960 US Pat. No. 6,654,150

  There are many conventional techniques based on the above-described principle, but there is a problem that in some cases, a color profile cannot be (preferably) obtained. That is, the test image includes not only a color sample portion in which a color sample (for example, a fruit basket image) is reflected but also a background portion (non-color sample portion) in which a landscape behind the color sample is reflected. When the lighting conditions for the color swatch portion and the non-color swatch portion are different from each other, or when the color swatch portion, the non-color swatch portion or both are excessively exposed, the color profile (preferably ) Can't ask. Further, even a relatively good conventional color profile generation method does not function properly unless the device-dependent data used for generating the color profile is raw 16-bit RGB data that has not been corrected. This is because, for example, clipping tends to occur in more general 8-bit image formats such as TIFF (registered trademark) and JPEG images. For this reason, the color profile is preferably used even when (a) the color sample portion is overexposed or underexposed compared to the non-color sample portion, and (b) the device-dependent data used is in a general 8-bit format. There is a need to be demanded.

  According to the digital image input device profiling system and method according to the present invention, the above-mentioned problems can be solved and a new technique can be brought to the industry. First, in one embodiment of the present invention, a color profile of a digital image input device is obtained based on at least an image of a color sample portion and estimated illuminance. The estimated illuminance used at that time is generated by comparing the coordinate value (illuminance) in the device-dependent color space of the color sample image with the coordinate value (illuminance) in the device-independent color space of the color sample. That is, since the color profile is obtained based on the illuminance estimated based on the data related to the color sample portion, according to the present embodiment, it is not necessary to refer to the data related to the non-color sample portion, that is, the background portion in the image when generating the color profile. . Therefore, according to the present embodiment, a color profile can be generated without being influenced by the relationship between the color sample partial exposure degree and the non-color sample partial exposure degree in the image. Furthermore, according to the system and method according to the present embodiment, a color profile can be suitably obtained even when clipping occurs in data. Therefore, the color sample image can be expressed in a general 8-bit image format.

  In one embodiment of the present invention, the scaling factor is obtained by comparing the coordinate value (illuminance) in the device-dependent color space and the coordinate value (illuminance) in the device-independent color space, and the color sample portion is obtained using the scaling factor. Linearly scale the coordinate values in the device-independent color space. Further, the optimum values of the parameter representing the gradation curve and the parameter representing the chromaticity in the color profile are obtained by calculation. Then, a more accurate color profile is obtained based on at least the device-independent color space coordinate values that have undergone linear scaling. Further, adjustment for each color with respect to the coordinate value in the device-independent color space is also useful for the accuracy of the color profile.

  Hereinafter, in order to contribute to a straightforward and accurate understanding of the present invention, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. As can be understood, the attached drawings are for explaining the idea of the present invention, and the dimensional ratios on the drawings do not necessarily match the actual ones.

  According to the present invention, various problems derived from underexposure or overexposure of the color sample portion with respect to the background portion (non-color sample portion) in the test image used when obtaining the color profile of the digital image input apparatus are alleviated or reduced. Can be solved. In particular, in the present invention, both the device-dependent data and the device-independent data relating to the color sample portion are used, and the background profile in the test image, that is, the device-dependent data acquired from the background is not used, and the color profile is determined. Can be sought. Therefore, the exposure of the background portion and the color sample portion is not affected. For this reason, in the following description, the present invention will be described based solely on the image data relating to the color sample portion, but the present invention can also be implemented in a form in which image data relating to the background portion is used together. Those skilled in the art (so-called persons skilled in the art) will understand that embodiments using both the color sample portion and the background portion in the image are not excluded from the present invention. Furthermore, the present invention can process clipped image data well, and can therefore be implemented based on image data in a general 8-bit format.

  FIG. 2 shows an input device profiling system 200 according to an embodiment of the present invention. The system 200 has a configuration in which a digital image input device 202 and a data storage system 206 are communicably connected to a computer system 204. The computer system 204 may be a system configured by a single computer, a system including a plurality of computers connected to each other through communication, or a configuration that requires an operation by the operator 208 or an unnecessary configuration. . In any case, the computer system 204 receives a test image captured by the input device 202 such as a digital camera or a scanner, that is, an image including a color sample portion (and other background portions that are not essential), and the test image. In addition, the color profile of the input device 202 is obtained based on the estimated illuminance of the color sample portion and the following procedure (see FIGS. 3 and 4), and the result is stored in the data storage system 206, or other device / Output in a form such as output to the computer 210. Information necessary for processing in the computer system 204, for example, information related to test images and color samples can be provided to the computer system 204 not only by the input device 202 but also by various means.

  The data storage system 206 is constituted by a computer memory, for example. The computer memory to be used, that is, the memory accessible from the computer may be one or plural. For example, the system 206 may be realized as a distributed data storage system including a plurality of computer memories that are communicably connected to a plurality of computers or apparatuses. One or a plurality of computer memories mounted on the computer or apparatus may be used as the system 206.

  Note that the term “computer” as used in the present application refers to all devices having a data processing function. Therefore, the computer in this application includes all devices suitable for data processing, management, handling, etc., such as desktop computers, laptop computers, mainframe computers, personal digital assistants (PDAs), Blackberry (registered trademark), etc. It is. The realization form may be any of electric, magnetic, optical, biological and so on.

  The term “computer memory” as used herein refers to all data storage / storage devices that can be accessed from a computer. Regardless of whether it is volatile or nonvolatile, it does not matter whether it is electrical, magnetic, optical or the like. Specific examples include a floppy disk, a hard disk, a compact disk, a DVD (registered trademark), a flash memory, a ROM, a RAM, and the like.

  Furthermore, the “communicable connection” in the present application refers to all connections that can be used for data communication between devices, computers, programs, etc., regardless of whether they are wired or wireless. Further, in the present application, “communication connection” includes connection between devices in one computer, between programs, etc., connection between devices incorporated in different computers, between programs, etc. There may be various forms such as connection between devices not incorporated in the device. Although the data storage system 206 is depicted separately from the computer system 204 in the drawings, some or all of the system 206 may be incorporated into the system 204 as will be understood by those skilled in the art from the foregoing description. May be.

  FIG. 3 illustrates an input device profiling procedure 300 according to one embodiment of the present invention. This procedure 300 is a procedure for obtaining a color profile of the digital image input apparatus 202, and is performed by hardware, software, firmware, or a combination thereof incorporated in the computer system 204.

  In step 302, computer system 204 receives device-dependent data representing a test image that includes at least a color swatch portion from an information source, such as digital image input device 202. In the present embodiment, since the RGB color space is assumed as the device-dependent color space, the device-dependent data is represented as RGB data in the following description. However, in the form of processing data complying with other types of device-dependent color spaces. The present invention can be implemented. In addition, the term “including at least the color sample portion” is used here because the test image may include a non-color sample portion, that is, a background portion. Therefore, the data received in step 302 includes the background data. The idea is that it may be. Furthermore, in the present application, the description will be made on the assumption that a color sample is used. However, any subject whose coordinate value in the device-independent color space corresponding to the color is known can be used as a color sample or a color sample. Can be used instead.

  The position and orientation of the color sample in the test image may be manually designated by an operator 208 operating the digital image input device 202, or may be automatically determined by the computer system 204, for example. In step 302, the data of the color sample portion of the received RGB data is averaged for each color patch to obtain a single RGB value representing the color patch for each color patch, and the device dependence of the array is determined. RGB data is output.

In step 304, RGB data 303, which is an RGB value array obtained from the input data, is coordinated in the device-independent XYZ color space indicating the true illuminance measured in advance for the color sample used in the test image. The data is stored as a two-dimensional data array RGB-XYZ 305 in association with XYZ data 301 as an array of values. The array RGB-XYZ 305 is composed of two data columns and generally a plurality of data rows. Of the data columns, one column stores device-dependent data corresponding to the data 303 in the figure, and the other column stores device-independent data corresponding to the data 301. Each data row is associated with any one color patch in the color sample portion, and data about the color patch is stored. Further, when data corresponding to the data 301 is stored in the array RGB-XYZ 305, the data may be converted from XYZ to LAB for the purpose described later. In this case, one of the data strings of the array RGB-XYZ 305 is device-dependent RGB data, and the other is device-independent LAB data. Information constituting such an array RGB-XYZ 305 is, for example, as follows on the software code:
An array that defines RGB data: rgbChartValue [i]
An array that defines LAB data: labMeasuredValue [i]
(Where 0 <i <numPatches; numPatches is the number of color patches included in the color sample portion). In this code, rgbChartValue [] is an RGB value array or RGB data representing each color patch in the input data, and labMeasuredValue [] is an array of LAB values of pre-measured true illuminance of the color patch included in the color sample portion. It is data. As can be understood by a so-called person skilled in the art, various methods other than this can be adopted as a method of associating the data 303 with the data 301.

  The data 301 in the figure is data conforming to the CIEXYZ color space, but any color space may be used as long as it is a device-independent color space. 3, 4, and 5 are examples in which the data 301 is compliant with the XYZ color space. However, the individual steps described in these drawings are different from the compliant color space of the data 301. This is also suitable for the device-independent color space. Therefore, although not shown in the figure because it is well known to those skilled in the art, those steps include processing for implicitly converting device-independent data from XYZ representation to other color space representations. I want to understand.

In step 306, the gray patch in the color sample portion is identified based on the LAB data in the two-dimensional data array RGB-XYZ 305, and the gray patch information 307 indicating the result is delivered to the subsequent stage. Further, the array RGB-XYZ 305 is not changed at this step and is transferred to the next step 308. Furthermore, as described above, the XYZ data 301 is converted to LAB and stored in the array RGB-XYZ 305. It is more advantageous to perform gray patch identification in step 306 in the LAB color space than in the XYZ color space. Because. That is, in the case of a gray color patch (gray patch), a and b of the coordinate values in the LAB color space are almost 0, so that the gray patch identification is considerably simplified by the conversion to LAB. Therefore, in step 306, gray patch identification is performed based on the LAB data in the array RGB-XYZ 305 (but this is not essential). The following software code [Equation 2]
grayIndex = 0
For (i = 0; i <numPatches; i ++)
{
If (labMeasuredValue [i] .a (0) <GrayRange && labMeasuredValue [i] .b (0) <GrayRange)
{
LabGrayValue.lab [grayIndex] = labMeasuredValue [i];
rgbGrayValues [grayIndex] = rgbChartValue [i];
grayIndex ++;
}
}
numGrayPatches = grayIndex;
Is the code for performing gray patch identification in such LAB color space. In this code, grayIndex is a work variable, i is a number for specifying a color patch to be processed, and GrayRange is a threshold value or a reference value for determining whether the color of the color patch is gray, for example, about 5ΔE. . As described above, the array rgbChartValue [] is the RGB data of the input image in the array RGB-XYZ305, and the array labMeasuredValue [] is the LAB data of the pre-measured true illuminance in the array RGB-XYZ305. A color patch in which the coordinate value a in the LAB color space given by i] .a (0) and the coordinate value b in the LAB color space given by labMeasuredValue [i] .b (0) are both less than the threshold GrayRange (here The gray patch can be identified by extracting the i-th color patch. The gray value of the identified gray patch in the CIELAB color space is stored and output in the array LabGrayValue.lab [], and the gray value of the gray patch in the RGB color space is stored and output in the array rgbGrayValues []. numGrayPatches is the total number of gray patches ultimately identified.

  As described above, in this embodiment, according to the above software code, the data or coordinate values related to the gray patch among the LAB data of the pre-measured true illuminance are set in the array LabGrayValue.lab [], and the corresponding input image RGB data or coordinate values are extracted and set in an array rgbGrayValues [] and output as gray patch information 307 to the subsequent stage. Further, it is convenient to sort the arrays LabGrayValue.lab [i] and rgbGrayValues [i] (where i = 0,..., NumGrayPatches) at this stage, and in order from light to dark. However, sorting is not necessary if it is known in advance that the order is correct.

In step 308, the in-limit brightest gray patch is identified from the gray patches, and the result is output to the subsequent stage as the brightest patch information 309. In the present embodiment, a device-dependent color space coordinate value that is identified as an in-limit brightest gray patch is within a limited range such as R <255, G <255, and B <255 (for an 8-bit image). Among the gray patches, the coordinate value in the LAB color space of the pre-measured true illuminance is the maximum (the illuminance is the highest). In other words, among the gray patches that are not overexposed when viewed from the estimated intensity in the RGB color space based on the input test image data, the brightest one as viewed from the coordinate value in the LAB color space of the pre-measured true illuminance. Identify as the lightest gray patch within limits. The RGB and LAB data required to perform step 308 are obtained from the two-dimensional data array RGB-XYZ 305. The gray patch information 307 and the array RGB-XYZ 305 are transferred to step 310 without any modification. In order to execute this step 308, for example, the following software code [Equation 3]
int maxGrayPatchIndex = 0;
float maxL = 0.0;
For (i = 0; i <numGrayPatches; i ++)
{
If (LabGrayValue.lab [i]> maxL && rgbGrayValues [i] .r () <255 && rgbGrayValues [i] .g () <255 && rgbGrayValues.b () <255)
{
maxL = LabGrayValue.lab [i];
maxGrayPatchIndex = i;
}
}
Should be executed. In this code, the gray patch information 307 and the array RGB-XYZ 305 given are already sorted according to the color coordinate value L or intensity in the LAB color space of the pre-measured true illuminance, in order from bright to dark. It is created on the premise of that. In this code, the array rgbGrayValues [] is an array of data for gray patches among the RGB data in the array RGB-XYZ 305, so rgbGrayValues [i] .r () is the R component intensity of the i-th gray patch, rgbGrayValues. Since [i] .g () is the G component intensity and rgbGrayValues [i] .b () is the B component intensity, their values for each gray patch and LabGrayValue.lab [], ie, in the array RGB-XYZ 305 By executing this code based on the coordinate values in the LAB color space, the position maxGrayPatchIndex of the in-limit brightest gray patch in the array LabGrayValue.lab [] and the color coordinate value L ^{* of the} patch maxL are obtained. Can do. The color coordinate value maxL is subsequently converted to Y _measured as described later.

In steps 310 and 312, in order to compensate for excessive illuminance or insufficient illuminance in the color sample portion, the data related to the color sample portion in the LAB data stored in the two-dimensional data array RGB-XYZ 305 is set to the illuminance correction coefficient α _ILCORRR . Use to scale. In this embodiment, the coefficient α _ILCORRR is obtained by obtaining the pre-measured true illuminance of the brightest gray patch within the limit expressed by the coordinate value in the LAB color space and the brightest gray patch within the limit expressed by the coordinate value in the RGB color space. It is generated by comparing the illuminance. That is, it is generated by comparing the pre-measured true illuminance for the lightest gray patch within the limits with the RGB color space coordinate value or illuminance for the same color patch obtained by the digital image input device 202, and The formula is [Equation 4]
α _ILLCORR = (Y _estimated ) / (Y _measured )
Is used. In this equation, Y _estimated is the estimated illuminance obtained from the RGB data for the within-limit brightest gray patch, and the following equation [5]
Y _estimated = f _Y (R _i0 , G _i0 , B _i0 )
It can ask for. The subscript i0 of each color component intensity is the in-limit lightest gray patch number represented by the variable maxGrayPatchIndex in the previous software code. Y _measured is LAB data of pre-measured true illuminance related to the brightest patch within the limit.

に示すように、右辺にあるデータベクトルＸＹＺの各成分値Ｘ，Ｙ，Ｚに係数α_{ＩＬＬＣＯＲＲ}を乗ずることによってそれらをＸＹＺ色空間内で照度補正し、左辺にあるベクトル即ち照度補正版装置ＸＹＺデータＸＹＺ_{ＩＬＬＣＯＲＲ}を生成する計算になる。限度内最明グレイパッチのＸＹＺ値とＲＧＢ値の対応関係は、この補正の前後で保存される。なお、ここでは、ガンマ値及びＲＧＢ色度がディジタル画像入力装置プロファイルデフォルト設定値に初期化されており（例えばガンマ値がカメラ向けの２．２に初期化されており）、当業界で知られているｓＲＧＢ（登録商標）乃至Ａｄｏｂｅ（登録商標）向けの値になっているものと、仮定している。そして、照度補正した装置非依存データを同じ色パッチに対応するＲＧＢデータを対応付けることによって、二次元データアレイＲＧＢ−ＸＹＺ_{ＩＬＬＣＯＲＲ}３１３を生成する。 In the present embodiment, the device-independent data in the two-dimensional data array RGB-XYZ305 is linearly scaled in accordance with the illuminance correction coefficient α _{ILCORRR in} step 312, and thereby the irradiance-corrected plate-independent data of the pre-measured true illuminance. Is generated. That is, if the device-independent data in the array RGB-XYZ ₃₀₅ is XYZ, for example, XYZ data XYZ _ILCORRR is generated, and if it is LAB, LAB data is generated. Taking XYZ as an example, this correction calculation is given by

As shown in FIG. 4, the component values X, Y, and Z of the data vector XYZ on the right side are multiplied by the coefficient α _ILCORRR to correct the illuminance in the XYZ color space, and the vector on the left side, that is, the illuminance correction version device XYZ data It is a calculation that generates XYZ _ILCORRR . The correspondence between the XYZ values and the RGB values of the brightest gray patch within the limit is saved before and after this correction. Here, the gamma value and the RGB chromaticity are initialized to the digital image input device profile default setting values (for example, the gamma value is initialized to 2.2 for the camera), and are known in the art. It is assumed that the values are for sRGB (registered trademark) to Adobe (registered trademark). Then, the two-dimensional data array RGB-XYZ _ILCORRR 313 is generated by associating RGB data corresponding to the same color patch with the device-independent data _subjected to illuminance correction.

  In step 314, a parameter list 315 suitable for executing step 316 is generated. That is, the parameter list 315 is generated so as to assist in the calculation of the final XYZ to LAB coordinate values in step 316 and, in turn, the derivation of the color profile 317 of the digital image input device 202 based thereon. The parameter list generation procedure in the present embodiment is as described with reference to FIG. 4, but other procedures known in the art can be applied to realize an alternative procedure. In step 316, RGB 303 and XYZ data 301 are used together with the parameter list 315 to generate a color profile 317 for the input device 202.

FIG. 4 shows the step 314 executed in the present embodiment in an exploded manner into a plurality of sub-steps. In step 402, a two-dimensional data array RGB-XYZ _ILCORRR 313 comprising RGB data as device-dependent data and XYZ data XYZ _{ILCORRR as} illuminance corrected device-independent data corresponding thereto is received, and a gradation curve description parameter α _TC 403 is generated. To do. The parameter α _TC 403 is one or a plurality of parameters representing a one-dimensional gradation curve of each color component of the device-dependent data in the array 313, that is, each of the RGB color components. In step 402, the value is optimized. In this step 402, the array 313 is not modified at all, and is passed to the step 404 as it is.

により表すことができる。この式では、色見本部分に含まれるグレイパッチの装置依存データＲＧＢ_ｉに対し、関数ベクトルF_{Ｌ＊ａ＊ｂ＊}（ＲＧＢ_ｉ，α_ＴＣ）によってパラメタα_ＴＣ４０３を適用し、その第ｉクレイパッチについてのＬＡＢ色空間内座標推測値を求める。更に、事前計測済真照度の照度補正版ＸＹＺベクトルＸＹＺ_{ＩＬＬＣＯＲＲ}をＬＡＢに変換したベクトルＬ^＊ａ^＊ｂ^＊ _{ＩＬＬＣＯＲＲ}から、求めたＬＡＢ色空間内座標推測値を減じて誤差を求める。そして、その自乗をｉ＝ｉ_０〜ｎ−１について総和することにより、自乗誤差総和関数Ｅｒｒ（α_ＴＣ）の値を求める。この式中、ｎは色見本部分に含まれるグレイパッチの個数、ｉはそれらに付した番号、ｉ_０はグレイ値のアレイLabGrayValue.lab[]における限度内最明グレイパッチの位置maxGrayPatchIndexを表している。なお、関数ベクトルF_ＬＡＢ（ＲＧＢ_ｉ，α_ＴＣ）に代わり当業界にて公知の他の関数を使用することもできる。そして、例えばＮｅｗｔｏｎ法やＰｏｗｅｌｌ法等の周知手順を使用し、自乗誤差総和関数Ｅｒｒ（α_ＴＣ）の値が最小になるようにパラメタα_ＴＣ４０３の値を自動的に変化させていくことにより、最適なパラメタα_ＴＣ４０３、即ち装置依存性のＲＧＢデータから装置非依存性のＸＹＺ乃至ＬＡＢ色空間内座標値を推測するのに適した値が得られる。 In the present embodiment, when obtaining and optimizing the gradation curve description parameter α _TC 403, first, the initial value of the parameter α _TC 403 is selected or specified, and applied to the RGB data to coordinate in the device-independent color space. A value, for example, a coordinate value in the XYZ to LAZ color space is estimated, and an error that the estimated value obtained thereby exhibits with respect to the coordinate value in the illuminance-corrected RGB color space of the pre-measured true illuminance is calculated. Furthermore, this procedure the error calculation result is the minimum (or sufficiently small; the same applies hereinafter) becomes to optimize the parameters alpha _{TC 403} by repeating while updating the parameter alpha _{TC 403,} the parameter alpha _{TC 403} at the time of minimum error Output to the subsequent stage. Accordingly, a part of the processing in step 402 of the present embodiment is as follows when LAB is used as the device-independent color space:

Can be represented by In this expression, the parameter α _TC 403 is applied to the device-dependent data RGB _i of the gray patch included in the color sample portion by the function vector F _{L * a * b *} (RGB _i , α _TC ), and the i-th clay The estimated coordinate value in the LAB color space for the patch is obtained. Further, an error is obtained by subtracting the obtained estimated coordinate value in the LAB color space from the vector L ^* a ^* b ^* _ILLCORR obtained by converting the illuminance corrected version XYZ vector XYZ _ILLCORR of the pre-measured true illuminance into LAB. Then, the value of the square error summation function Err (α _TC ) is obtained by summing the squares for i = i _{0 to} n−1. In this equation, n is the number of gray patches included in the color sample portion, i is the number assigned to them, and i ₀ is the position of the brightest gray patch within the limit, maxGrayPatchIndex, in the array of gray values LabGrayValue.lab []. Yes. Note that other functions known in the art may be used instead of the function vector F _LAB (RGB _i , α _TC ). Then, for example, by using a known procedure such as the Newton method or the Powell method, the value of the parameter α _TC 403 is automatically changed so that the value of the square error summation function Err (α _TC ) is minimized. The optimum parameter α _TC 403, that is, a value suitable for estimating the device-independent XYZ to LAB color space coordinate values from the device-dependent RGB data is obtained.

As the function vector F _{L * a * b *} (RGB _i , α _TC ), for example, the following equation [Equation 6]
_{f R (R, R Max,} R bias, γ R, β C) = R Max (1.0-R bias) [f C (R, β C)] γR + R bias
The function f _R represented by can be used. This function f _R is a function of the R component, the function of the same shape can be used also for the G and B components. The function f _R is a function that gives a one-dimensional gradation (response) curve for the R component. When the digital image input device 202 to be used is, for example, a digital camera, the gradation curve description parameter α _TC is expressed as shown in this equation. R _Max , R _bias , γ _R and β _C are used. The function in the formula [Equation 7]
f _C (x, β _C ) = x + β _C (0.5 + 0.5 (2 (x−0.5)) ³ −x)
Is a contrast function with RGB color component intensities x and parameter β _C as arguments. When β _C = 0, f _C (x) = x. Further, R _Max of the tone curve description parameter α _TC is the maximum intensity of the R component obtained by the input device 202 to be used, and is used as a linear scaling coefficient. R _bias is the minimum intensity of the R component obtained by the input device 202 and is used as a black bias offset. γ _R is a gamma value of the input device 202, particularly a gamma value for the R component, and the value is as known in the art. When calculating the G component and the B component, the gamma value γ _G for the G component and the gamma value γ _B for the B component (in the same order) are used, but the input device is used as the gamma values γ _R , γ _G and γ _B. A total gamma value of 202, that is, the same value may be used. β _C is a contrast adjustment parameter, and the value of the contrast function f _C is decreased in the range of 0.0 <x <0.5, and is increased in the range of 0.5 <x <1.0. Here, a cubic function of x is used as the contrast function f _C , but there are various functions such as a spline function as a mathematical function that can perform the same adjustment. This kind of adjustment itself is a well-known idea in the industry including an automatic contrast function incorporated in an application such as Adobe (registered trademark) PhotoShop (registered trademark). It was intended to improve the aesthetics of the device and was not intended to describe and evaluate the color characteristics of the device.

The step 402 in the present embodiment is, for example, the following function on the software code:
MinimizeError (sumSquareGrayDeltaE, parameterList, NumParameters)
This can be realized by executing This function is a function that minimizes the return value of the function sumSquareGrayDeltaE () and returns the variable parameterList and NumParameters, that is, a function that obtains parameterList and NumParameters that minimizes the return value of the function sumSquareGrayDeltaE (). If all gray patches are sorted and arranged in order from light to dark, the code for the function sumSquareGrayDeltaE () is for example [Equation 9]
For (i = brigetestValidGray; i <nGray; i ++)
{
xyzPred = evaluateModel (grayRGB [i]);
labPred = mColMetric-> XYZToLab (xyzPred);
labMeas = grayLab [i];
xyzMeas = mColMetric-> LabToXYZ (labMeas);
xyzMeas * = Y_IlluminationCorr;
labMeas = mColMetric-> XYZToLab (xyzMeas);
errSq = distanceSquaredLab (labPred, labMeas);
sumSq + = errSq;
}
return (sumSq);
It becomes. In this code, brightestValidGray represents the position of the brightest gray patch within the limit in the illuminance correction _plane device-independent data (XYZ _{ILCORRR in} this example), and nGray represents the total number of gray patches. xyzPred is generated by the function evaluateModel () that encodes the function vector F _LAB (RGB _i , α _TC ) described above, and coordinates in the XYZ color space corresponding to the coordinate values in the RGB color space in the device-dependent data 303. It represents an estimated value. In this code, labPred is obtained by converting the compliant color space of the estimated value xyzPred from XYZ to LAB using the function mColMetric-> XYZToLab (). Also, labMeas represents the coordinate value grayLab [] in the LAB color space of the current gray patch (that is, the i-th gray patch). In this code, xyzMeas is obtained by converting the compliant color space of the coordinate value labMeas from LAB to XYZ, and xyzMeas ^* is obtained by correcting xyzMeas with Y_IlluminationCorr representing the illuminance correction coefficient α _ILCORRR . errSq is the square error of labPred for each labMeas, and sumSq is the sum. ParameterList is a variable for outputting the optimized gradation curve description parameter α _TC 403, and NumParameters is a variable for outputting the number of parameters.

The gradation curve description parameter α _TC 403 needs to be adjusted so that the total sumSq of the calculated square error errSq is minimized. A mathematical model for estimating the coordinate values in the XYZ to LAB color spaces corresponding to the coordinate values in the RGB color space in the two-dimensional data array 313, such as the function evaluateModel (), is known in the art. A model that is as smooth as possible (a model with a sufficiently low second derivative value, or unnecessary visual banding (strip defects) when the image is adjusted based on the obtained color profile). It is desirable to use a model that does not cause false images, or a model that can perform accurate estimation with as few parameters as possible. The following software code [Equation 10]
double evalGammaModel (double * RGBMax, double * gamma, double * blackBias, double filmContrastCorr, int rgbIndex, const double & cVal)
{
double val, corrVal, result = 0.0;
corrVal = 0.5 + 0.5 * pow (2.0 * (cVal-0.5), 3.0) -cVal;
val = cVal + filmContrastCorr * corrVal;
If (rgbIndex! = IFixedMaxRGB)
result = RGBMax [rgbIndex] * (1.0-blackBias [rgbIndex]) * pow (val, gamma [rgbIndex]) + blacBias [rgbIndex];
else
result = (1.0-blackBias [rgbIndex]) * pow (val, gamma [rgbIndex]) + blacBias [rgbIndex];
if (cVal <0.0)
result = 0.0;
return (result);
}
Is a function evalGammaModel () used as a function vector F _LAB (RGB _i , α _TC ), that is, a mathematical model used in the present embodiment for estimating a coordinate value in a device-independent color space based on RGB data. It is a written code. Among the parameters used in this function, those corresponding to the parameter α _TC 403 are RGBMax [], gamma [], blackBias [], and filmContrastCorr. Of these, RGBMax [] represents the RGB component maximum intensities R _Max , G _Max, and B _Max obtained from the digital image input device 202 and is used as a linear scaling coefficient. gamma [] represents the gamma values γ _R , γ _G and γ _B of each RGB component. blackBias [] represents the RGB component minimum strengths R _bias , G _bias, and B _bias obtained from the input device 202, and is used as a black bias offset. filmContrastCorr corresponds to contrast adjustment parameter beta _C. The argument rgbIndex is an argument indicating which color is currently being calculated. If the value is 0, the R component, 1 is the G component, and 2 is the B component. . Furthermore, the variable cVal is assigned to the function evalGammaModel () and given RGB value to be converted into coordinate values in the color space by the function evalGammaModel (), the variable corrVal is the contrast function value, and the variable result is rgbIndex = 0,1,2 This is a variable for outputting each RGB component value obtained for. The RGB vector represented by the output variable “result” is converted into an XYZ vector by multiplication with an RGB → XYZ conversion matrix as described later. In the RGB → XYZ conversion, the above-mentioned square error sum calculation is repeatedly executed while adjusting the parameter α _TC 403 according to a predetermined procedure until the square error sum sumSq is minimized.

More specifically, in the calculation using the mathematical model, an RGB value correction amount corrVal is first generated by contrast adjustment calculation for a given RGB value cVal. The formula used for this is a cubic polynomial, and when the given value cVal is equal to 0, 0.50 or 1.0, the correction amount corrVal is equal to the given value cVal (the RGB color space coordinate value is normalized to 1.0 ). The generated correction amount corrVal is multiplied by the contrast adjustment parameter filmContrastCorr, and the result is added to the given value cVal, thereby obtaining the RGB component values val of the corrected version. The default value of the parameter filmContrastCorr is 0.0, but the value can be positive (that is, reduction correction) or negative (that is, increase correction). The value val obtained in this way is the standard gamma function calculation for the black bias offset blackBias [], the linear scaling coefficient RGBMax [] and the gamma value gamma [] of each RGB component as the estimated coordinate value xyzPred in the XYZ color space, Used in calculations to adjust the curve description parameter α _TC 403. In this example, there are ten types of parameters α _TC 403 to be adjusted. That is, all are the linear scaling coefficient RGBMax [], the gamma value gamma [] and the black bias offset blackBias [] for each RGB component, and the contrast adjustment parameter filmContrastCorr common to each RGB component, but the gamma value gamma [] If common among the RGB components, the parameter α _TC 403 to be adjusted is reduced to eight types. This method is useful for a low-quality image in which a relatively large amount of noise is included in coordinate values in a device-dependent color space (for example, RGB color space) extracted from the color sample portion. Further, the gradation curve calculation is repeated according to the number of gray patches included in the color sample portion. For example, if the number of gray patches nGray in the color sample portion is 6, the number of data points to be calculated is 6 patches × 3 coordinate values (L ^* , a ^* , b ^* ) = 18 points. If two of the gray patches in the color sample portion are overexposed, they are excluded from the calculation target, and the number of calculation target data points is reduced to 4 patches × 3 coordinate values = 12 points.

After obtaining the optimum parameter α _TC 403 in step 402, an optimum chromaticity description parameter α _C 405 representing the RGB chromaticity on the color profile of the digital image input device 202 is generated in step 404. The parameter α _TC 403 and the two-dimensional data array 313 are passed to step 406 without any modification. As will be described later, in step 404, an RGB color space is used for all data in the color sample portion in accordance with a standard matrix / tone curve (TRC) arithmetic expression and using the gradation response calculated in step 402. For example, a coordinate value in the XYZ color space in the device-independent color space corresponding to the internal coordinate value is calculated. As the error function, the same one as described with respect to step 402 is used, except that the calculation is performed on all data, and for example, when any of the RGB values is 0 or 255, a predetermined value is used. The square error is reduced according to the weighting factor and added to the total value. This is a process that takes into account that in the case of image data that has undergone clipping, actual coordinate values in the device-dependent color space, such as RGB color component values, cannot be accurately determined.

によって数学的に記述される単純な関数になる。即ち、ＲＧＢ各成分値はこのマトリクスＭを用いＸＹＺ各成分値に容易に変換することができる。このマトリクスＭはＲＧＢ各成分の色度（ｘ，ｙ）、並びにそのシステム乃至モデルのホワイトポイント即ちＲＧＢ各成分値が最大であるときのＸＹＺ各成分値から求めた色度（ｘ，ｙ）によって、次の式

の通りユニークに決まる。この式中の添え字ｒ_ｉ，ｇ_ｉ，ｂ_ｉはＲＧＢ各成分を、添え字ｗｐはホワイトポイントをそれぞれ表している。また、この場合に用いる誤差関数は、（ａ）グレイパッチだけでなく色見本部分内の全てのデータを使用すること、並びに（ｂ）パラメタα_ＣがＲＧＢ各成分毎に（ｘ，ｙ）値として定義されていることを除いて、ステップ４０２にて定義された階調曲線のそれと同様の形式になる。更に、このシステム乃至モデルのホワイトポイント（ｘ_ｗｐ，ｙ_ｗｐ）は、当業界で知られている色見本計測データ例えばＤ５０によるデータ向けのホワイトポイントに相当するものにすべきである。 The chromaticity description parameter α _C 405 adjusted and optimized in step 404 is, for example, a chromaticity value for each of the RGB components, that is, a value expressed as (x, y) in the industry. As described above, the RGB data in the two-dimensional data array 313 is processed by the gradation curve function in step 402, and the RGB component values are converted into values in the linear color space. Independent color space coordinate values, for example, XYZ component values are converted into the following matrix

Makes a simple function mathematically described. That is, RGB component values can be easily converted into XYZ component values using this matrix M. This matrix M is based on the chromaticity (x, y) of each RGB component and the chromaticity (x, y) obtained from the white point of the system or model, that is, the XYZ component values when the RGB component values are maximum. And the following formula

It is uniquely determined as follows. Subscript _r i in this _equation, g i, the _{b i} for each RGB component, subscript wp represents respectively the white point. The error function used in this case is (a) not only the gray patch but also all data in the color sample part, and (b) the parameter α _C is an (x, y) value for each RGB component. Is the same format as that of the gradation curve defined in step 402. Furthermore, the white point (x _wp , y _wp ) of this system or model should correspond to the white point for color sample measurement data known in the art, for example data for D50.

In step 406, an optimum color-specific adjustment amount description parameter α _SA 407 is generated (however, this step is not essential). The parameter α _SA 407 is a parameter that describes an adjustment amount for each color with respect to coordinate values in the device-independent color space generated in step 404, for example, XYZ color component values, for the color profile of the digital image input device 202. As a routine for obtaining the optimum parameter α _SA 407, for example, an error minimizing routine similar to that already described with respect to step 402 and step 404 may be used. When an image is converted or rendered in accordance with the color profile 317 obtained by the processing according to the present embodiment, a slight image or unnecessary deterioration may occur in the obtained image. By appropriately determining the parameter α _SA 407 for one type or a plurality of types of colors, and using the parameter α _SA 407 to adjust the coordinate values in the device-independent color space for each color, such possibility or danger can be suppressed to a considerable extent. . Various procedures can be used to determine the adjustment amount for the processing in step 406, that is, the coordinate values in the device-independent color space. For example, it is desirable to use the procedure shown in FIG. In the procedure illustrated in FIG. 5, a group of color adjustment amounts ΔXYZ _i is used to generate the parameter α _SA 407 (i = 0,..., 5; each i is R, G, B, C, M, Y), and by optimizing them using the error minimization procedure as described above, the adjusted version XYZ color space coordinate value XYZ _ADJ 505 that minimizes the square error sum is generated.

In step 408, the gradation curve description parameter α _TC 403, the chromaticity description parameter α _C 405, and the color-specific adjustment amount description parameter α _SA 407 or a part thereof are finally optimized (however, this step is not essential). . This optimization can be performed for each parameter, for example, by re-executing Steps 402, 404, or 406, or can be performed collectively using the error minimization procedure described above, for example, the Powell method. In the case of an image that is not overexposed, the number of calculation target data points in this optimization is, for example, 24 × 3 = 72 points. The parameters to be optimized are, for example, 10 types of parameters α _TC 403, 6 types of parameters α _C 405, and 18 types of parameters α _SA 407. Further, since there is a strong correlation between the chromaticity value and the coordinate value adjustment amount in the device-independent color space, for example, the XYZ color space, the chromaticity is fixed and the global optimization for the gradation curve and the XYZ color are performed. It is desirable to perform global optimization on the coordinate value adjustment amount in space simultaneously. The parameters optimized in step 408, that is, the chromaticity description parameter α _C '409 of the adjusted version, the tone curve description parameter α _TC ' 410, and the color-specific adjustment amount description parameter α _SA '411 are the illumination correction coefficient α _It is included in the parameter list 315 together with the _ILLCORR 311 and used for the _subsequent processing, that is, the generation processing of the color profile 317 in step 316. The two-dimensional data array 313 does not need to be transferred to the subsequent stage. Further, the processing for generating the array 313 (step 312 in FIG. 3) can be executed in step 316. In this case, since the array 313 is not obtained prior to the execution of step 314, step 314 is executed. To generate it. That is, by _{applying the} coefficient α _ILLCORR 311 obtained in step 310 to the XYZ to LAB data in the two-dimensional data array RGB-XYZ305 delivered from step 310, the illuminance correction version XYZ data XYZ _ILLCORR (or correspondingly) LAB data) and thus the array 313 containing it is generated at an appropriate stage in step 314.

  According to the procedure shown in FIGS. 3 and 4, a color sample known in the art, for example, a color checker portion is overexposed or underexposed compared to other portions of the image, that is, the background portion. Even in this case, a favorable result can be obtained. In particular, the remarkably strong color and the white balance of the entire image are improved without exception. Further, by correcting the value of the pre-measured true illuminance XYZ data 301, it is possible to realize a color profile equivalent to a so-called look profile, for example, to saturate an image more.

  In turn, FIG. 5 shows an apparatus independent color space coordinate value adjustment procedure 500 in this embodiment. According to this procedure 500, the appearance of the visually displayed image and the appearance of the hard copy output image can be suitably matched. That is, this procedure 500 is useful for examining the characteristics of image data obtained from the digital image input device 202, rendering the image on a display device, and converting the image for a digital image output device. It is also useful in obtaining an accurate color profile 307 for the input device 202, such as a digital camera or scanner. Therefore, in the present embodiment, the procedure 500 is executed in step 406 in FIG. 4. However, as can be understood by a person skilled in the art, the procedure 500 is completely independent from the procedure 300 shown in FIGS. 3 and 4. It can also be implemented in the form.

  In this procedure 500, first, a correction coefficient is derived based on device-dependent data representing an input image, and the correction coefficient is applied to the entire region or a partial region of the device-dependent data, thereby substantially generating a false image. The device-independent data is linearly corrected so as not to occur. The correction for the device-independent data is adjustment for each color, which is useful for improving the color in the designated color region. The target color area is specified along each axis of RGBCMY, for example, and color deterioration does not occur in other color areas.

Information used when executing this procedure 500 is represented by RGB data 501 as device-dependent data, XYZ data 502 as device-independent data, other parameters 503, and coordinate values in the device-independent color space, for example, XYZ color space. The color-specific adjustment amount ΔXYZ _i 504. i is a subscript representing each color region. When this procedure 500 is executed as part of step 406 in FIG. 4, the color-specific adjustment amount ΔXYZ _i 504 is set using the gradation curve description parameter α _SA 407. Further, the data 501 and 502 are data having a correspondence relationship between the data fragments included in both, and when this procedure 500 is executed as part of step 406 in FIG. 4, the data in the two-dimensional data array 313 is stored. The RGB data is used as data 501, and the corresponding XYZ data (or LAB data) XYZ _ILLCORR in the same array 313 is used as data 502.

In order to improve the color profile of the RGB data 501 obtained from the digital image input device 202, correction is performed in a device-independent color space similar to that obtained from the characteristic expression of the device relating to the data. desirable. In addition, the characteristic notation of the RGB data generation device such as a digital camera or a scanner includes the matrix / TRC arithmetic expression described with respect to step 402 in FIG. 4 and similar expressions, that is, conversion from the RGB color space to the XYZ color space. An expression can be used. Therefore, in most apparatuses, correction can be suitably performed in the XYZ tristimulus color space. Therefore, in this embodiment, the device-independent data 502 to be adjusted and the device-independent data XYZ _ADJ 505 after adjustment are both coordinate values in the XYZ color space. However, the linear correction according to this procedure 500 is not limited to the coordinate values in the XYZ color space, but also to various CIECAM models known in the art, such as CIELAB, such as CIELAB, and also to a certain extent. Can be applied successfully.

A color-specific adjustment amount ΔXYZ _i 504 represents a correction amount for a specified color region i in the XYZ color space, for example, a correction amount for saturation, tone, and luminance in the region i. In the present embodiment, the value range of the argument i indicating the designated color region is, for example, 0 ≦ i ≦ 5, and each RGBCMY color component value is associated with i = 0 to 5 in order, R ₀ , G ₀ , B ₀ , C _0. , M ₀ , Y ₀ . For example, the color of R = R _max , G = O, B = O, C = O, M = O and Y = O is expressed as R ₀ (G ₀ , B ₀ , C ₀ , M ₀ and Y ₀ are also the same). . As can be understood by a so-called person skilled in the art, it is possible to specify a color region in other forms. The method for generating the color-specific adjustment amount ΔXYZ _i 504 may be manual generation or automatic generation. In the case of manual generation, for example, a procedure may be used in which an image is displayed in a CRT display or hard copied based on the XYZ data 502, the operator visually confirms the image, and specifies a color to be corrected or adjusted. In this procedure, for example, if the R component in the displayed image is too bright, the operator who senses it makes the color adjustment amount ΔXYZ _{0 for} i = 0 negative, that is, the luminance value of the R component is negative, The luminance of the R component is reduced. In the case of automatic generation, for example, the error minimizing routine described in relation to step 402 and step 404 may be used. For example, when this procedure 500 is executed as part of step 406 in FIG. 4, first, the initial value of the color-specific adjustment amount ΔXYZ _i 504 is automatically set, and then the value of the color-specific adjustment amount ΔXYZ _i 504 is changed in various ways. On the other hand, the routine according to this procedure 500 is repeatedly executed to obtain, for example, the adjusted version XYZ data XYZ _ADJ 505 so that the sum of square errors with respect to the device-independent color space coordinate estimated values xyzPred to labPred is minimized. When the sum of square errors is minimized, the value of the color-specific adjustment amount ΔXYZ _i 504 at that time is used as a gradation curve description parameter α _SA 407 for processing subsequent to step 406.

In this procedure 500, in step 506, linear RGB data 507 (RGB) ′ is generated based on the RGB data 501 and other parameters 503. If the data 501 is the data 507 itself, the acquisition of other parameters 503 and the execution of step 506 can be omitted. The generation procedure of the data 507 is arbitrary, but in principle, the data that can optimize the estimated value of the XYZ data 502 by multiplying the appropriate RGB → XYZ conversion matrix as shown in the optimization procedure described above. 507 is generated by calculation for correcting the data 501. In the case of a CRT, the data 507 is calculated according to a gamma curve relating to the CRT, for example, ^Rγ . In this case, as other parameters 503, parameters describing the gamma curve are used. When higher accuracy is required or when it is used for an LCD system, it is necessary to use a more complicated function than just a gamma curve. Furthermore, when this procedure 500 is executed in step 406 in FIG. 4, one or more of the gradation curve description parameters α _TC 403 are used as (part of) other parameters 503. For example, when adjusting the coordinate value in the device-independent color space representing the color profile of a digital camera, in step 506, functions f _R , f _G , f _B (and step 402 representing the gradation curve of the camera) are adjusted. Data 507 is generated using the corresponding parameter α _TC 403) obtained in the above.

In step 508, a color-specific adjustment coefficient β _i 509 is generated for each color region i corresponding to the color-specific adjustment amount ΔXYZ _i 504. That is, the coefficient β _i 509 is calculated using an individual linear correction function. Each individual linear correction function exhibits a maximum value at a boundary portion of a corresponding device-dependent color space, that is, a compliant color space (RGB color space) of the RGB data 501, and the value of the data 501 is another boundary portion or a medium portion. When the vertical axis is approached, the value decreases linearly and becomes zero or almost zero. In other words, the calculation of the coefficient β _i 509 in this embodiment is as follows: (a) The coordinate value in the device-dependent RGB color space that is being investigated is the maximum from all other colors related to the designated color region i and the neutral axis. When the distance is, the value of the individual linear correction function for the region i and the coefficient β _i 509 is the maximum value, and (b) the coordinate value in the RGB color space is related to the designated color region i and other neutral axes. As the color is approaching, the individual linear correction function values for that region i and coefficients β _i 509 are linearly approaching zero. In order to capture the meaning of “related” here, it is preferable to assume some boundary color R _b G _b B _b . For example, if each color component value of the boundary color R _b G _b B _b is either 0 or 1.0, the other color “related” to the boundary color R _b G _b B _b is, for example, the boundary color R _b G _b B A color adjacent to _{b b} , that is, a color obtained by changing any one of the color component values from 0 to 1 or from 1 to 0. Therefore, the other colors related to R _b G _b B _b = (1, 0, 0) are (1, 1, 0) and (1, 0, 1), and R _b G _b B _b = (1, The other colors related to (1, 0) are (1, 0, 0) and (0, 1, 0).

に従い計算する。但し、ここではＲＧＢ各色成分値が１．０に正規化されており、従って係数β_ｉ５０９の値域が０〜１の範囲であり、ひいてはβ_ｉ＝１にて補正量が最大になるものとする。例えばｉ＝０即ちＲ成分に係る係数β_０は、いま調べている装置依存色空間内座標値即ちＲＧＢデータ５０７の値が、その色領域ｉに関わる他の全ての色即ちＧＢＣＭＹの各色成分値及び中立軸から最大距離にあるときに、最大値になる。言い換えれば、データ５０７のうちいま調べているＲＧＢ色空間内座標値のうちのＧ成分及びＢ成分が０であると判明したときに、係数β_０は最大になる。また、データ５０７のうちいま調べているＲＧＢ色空間内座標値がＧＢＣＭＹの何れかの成分値又は中立軸に接近するにつれ、係数β_０に対応する個別的線形補正関数はその係数β_０を線形的に０にスケーリングする。言い換えれば、データ５０７のうちいま調べているＲＧＢ色空間内座標値のうちのＧ成分及びＢ成分が増大していることが判明したときに、個別的線形補正関数は係数β_０を線形的に０にスケーリングする。 In this embodiment, the color adjustment coefficient β _i 509 for each i = 0 to 5 corresponds to the value β _i , that is, R ₀ , G ₀ , B ₀ , C ₀ , M ₀ and Y ₀ in order. A value to be multiplied by the corresponding one of the color-specific adjustment amounts ΔXYZ _i 504 is expressed by the following equation:

Calculate according to However, here, the RGB color component values are normalized to 1.0, and therefore the value range of the coefficient β _i 509 is in the range of 0 to 1, and therefore the correction amount is maximized when β _i = 1. To do. For example, i = 0, that is, the coefficient β ₀ relating to the R component is the coordinate value in the device-dependent color space being examined, that is, the value of the RGB data 507 is the value of all other colors relating to the color region i, that is, the color component values of GBCMY. And when it is at the maximum distance from the neutral axis, it reaches the maximum value. In other words, when it is determined that the G component and the B component of the coordinate values in the RGB color space that are currently examined in the data 507 are 0, the coefficient β ₀ is maximized. Further, as the RGB color space coordinate values are investigating now among the data 507 is close to one of the component values or neutral axis of the GBCMY, the individual linear correction function corresponding to the coefficient beta ₀ the coefficient beta ₀ Linear Scaling to zero. In other words, when it is found that the G component and the B component of the RGB color space coordinate values currently examined in the data 507 are increased, the individual linear correction function linearly sets the coefficient β _0. Scale to zero.

により加重総和として計算する。 In step 510, the individual adjustment amount ΔXYZ _i 504 for each color in the XYZ color space is corrected based on the corresponding adjustment factor β _i 509 for each color, and the sum, that is, the adjustment amount ΔXYZ _TOTAL 511 common to each color for the XYZ data 502 is calculated. Is done. The common color adjustment amount ΔXYZ _TOTAL 511 in the device-independent color space is the sum of the individual color adjustment amounts ΔXYZ _i 504, for example,

To calculate as a weighted sum.

In step 512, the XYZ data 502 is adjusted based on the color common adjustment amount ΔXYZ _TOTAL 511 in the XYZ color space, and adjusted XYZ data XYZ _ADJ 505 is generated. For example, the following equation [Equation 11]
XYZ _ADJ = XYZ + ΔXYZ _TOTAL
Data XYZ _ADJ 505 is calculated according to

となる。 The software code for executing this procedure 500 is, for example,

It becomes.

According to this procedure 500, soft proofing on the display screen can be performed in a more suitable procedure. To do this, first correct the color profile A of the display to a better color profile A ′ and then convert the corrected color profile A ′ to the original color profile A to see what the color profile is. What is necessary is just to estimate how much should be changed. For example, when the operator determines that it is better to shift the Y color tone in the displayed image in the R direction, the saturated Y is shifted in the R direction by 3ΔE as the Y component correction amount ΔXYZ ₅ . The procedure 500 may be performed using the value. When the operator confirms that the desired result is obtained, the color profile of the display in the RGB color space is corrected with the reverse tendency so that the desired R-direction Y tone shift appears in the image rendered on the display screen. To. That is, the Y component correction amount DerutaXYZ ₅ previously used to adjust the color profile A display by Y component correction amount DerutaXYZ ₅ of opposite polarities. This can be done with sufficiently high accuracy.

  Similarly, the coordinate values in the device-independent color space can be corrected using this procedure 500. For example, the original color profile is converted from the color profile based on the XYZ color space to the color profile A1 based on the RGB color space, corrected by the procedure 500 to generate a new color profile A2, and the color profile A2 May be converted from the RGB color space to the XYZ color space. That is, in general terms, an XYZ color space-compliant color profile can be corrected along a conversion path of XYZ → RGB (profile A1) → RGB (profile A2) → (XYZ) ′. As described above, the process of jointly executing a plurality of types of conversions can be referred to as profile connection. The color profile obtained by the XYZ → (XYZ) ′ conversion can be called an abstract profile.

Still another application of this procedure 500 is the application described with reference to FIGS. 3 and 4, that is, a process for obtaining a color profile of the digital image input device 202 such as a scanner or a digital camera. According to this procedure 500, by correcting the general color profile of a digital camera or scanner expressed in the RGB color space appropriately, the gray balance is ensured and the integrity and consistency of the entire image are increased. It is possible to maintain the standard, and it is possible to suitably perform speculative conversion from RGB to XYZ or L ^* a ^* b ^* while maintaining the color of the subject in the image. The values of angle, saturation, and luminance correction amount for each color of RGBCMY can be automatically determined by an error minimizing routine.

Unlike the conventional method, since the individual linear correction function is used in this procedure 500, for example, there is a relatively small color difference of the order of 2 to 3ΔE, and a relatively large color difference of the order of 20 to 30ΔE. Can be corrected. In addition, the example of a structure of this invention is shown as an additional note below.
(Appendix 1)
A method for obtaining a color profile of a digital image input device by executing profiling based on a color sample portion of a color sample image and its estimated illuminance without using the relationship between the color sample portion exposure and the non-color sample portion exposure. .
(Appendix 2)
The method according to appendix 1, wherein the profiling is performed to check the coordinate values in the device-dependent color space in the color sample portion.
(Appendix 3)
The method according to appendix 2, wherein the estimated illuminance is obtained from the coordinate values in the device-dependent color space that have been examined.
(Appendix 4)
The method described in Appendix 2 for profiling,
Linearly scaling the device-independent color space coordinates in the color swatch using the scaling factor,
A method of calculating a first parameter representing a gradation curve of the input device based on the linearly scaled device-independent color space coordinate value.
(Appendix 5)
The method according to appendix 4, wherein the ratio of the estimated illuminance to the true illuminance in the color sample portion is used as the scaling factor.
(Appendix 6)
The method according to claim 5, wherein when the estimated illuminance is higher than the true illuminance, the linear scaling is performed so that the coordinate value in the device-independent color space increases linearly.
(Appendix 7)
The method according to claim 5, wherein when the estimated illuminance is lower than the true illuminance, the linear scaling is performed so that the coordinate value in the device-independent color space is linearly reduced.
(Appendix 8)
The method according to appendix 4, wherein a second parameter representing chromaticity in a color profile is calculated based on the linearly scaled device-independent color space coordinate values in profiling.
(Appendix 9)
The method according to appendix 1, wherein the illuminance of the brightest gray patch among a plurality of color patches included in the color sample portion is estimated and the result is used as the estimated illuminance.
(Appendix 10)
The method according to appendix 1, wherein the image is an 8-bit image.
(Appendix 11)
A method for obtaining a color profile of a digital image input device, comprising:
Capturing a color sample image including a plurality of color patches with the input device;
Examining the coordinate values in the device-dependent color space of each color patch;
Estimating the illuminance of the brightest gray patch among the color patches based on the investigated device-dependent color space coordinate values;
Linearly scaling the coordinate values in the device-independent color space of each color patch based on the obtained estimated illuminance;
Determining a color profile of the input device based on the linearly scaled device-independent color space coordinate value or a derived amount thereof;
Storing the obtained color profile in a computer memory;
Having a method.
(Appendix 12)
The method according to claim 11, wherein the color profile is obtained without using the relationship between the color sample partial exposure and the non-color sample partial exposure in the image.
(Appendix 13)
The method according to claim 11, further comprising the step of identifying a gray patch in the color sample portion based on coordinate values in the device-independent color space of each color patch.
(Appendix 14)
The method according to attachment 13, wherein the gray patch is identified,
Compare the coordinate value in the device-independent color space of each color patch with a predetermined upper limit gray value,
A method of identifying a gray patch whose coordinate value in the device-independent color space is less than the upper limit gray value among color patches.
(Appendix 15)
15. The method according to appendix 14, wherein the upper limit gray value is about 5ΔE.
(Appendix 16)
The method according to claim 11, wherein linear scaling with respect to the coordinate value in the device-independent color space of each color patch is determined by calculating a ratio between the true illuminance and the estimated illuminance for the brightest gray patch among the color patches. To do based on the scaling factor.
(Appendix 17)
The method according to claim 16, wherein the true illuminance is obtained based on the coordinate value in the device-independent color space of the brightest gray patch.
(Appendix 18)
The method according to appendix 11, wherein
Calculating a first parameter representing a tone curve of the input device based on the linearly scaled device-independent color space coordinate values;
Adjusting the linearly scaled device-independent color space coordinate value or the amount derived therefrom using a separate linear correction function;
And determining the color profile of the input device using the coordinate value or derivative amount in the device-independent color space adjusted for each color and the first parameter.
(Appendix 19)
The method according to appendix 18, wherein the first parameter is calculated by using each device-dependent color space coordinate value estimation parameter, and based on the device-dependent color space coordinate value, each gray included in the color sample portion. A method for obtaining an estimated value of coordinates in a device-independent color space of a patch.
(Appendix 20)
The method according to appendix 19, wherein, for each gray patch included in the color sample portion, the difference between the device-independent color space coordinate estimation value with respect to the linearly scaled device-independent color space coordinate value is minimized. The method for adjusting the parameter for estimating the coordinate value in the device-dependent color space.
(Appendix 21)
The method according to appendix 19, wherein all use the black bias offset, the linear scaling coefficient, and gamma given for each color axis of the device-dependent color space as the parameter for estimating the coordinate value in the device-dependent color space.
(Appendix 22)
The method according to appendix 19, wherein as the device-dependent color space coordinate value estimation parameter, a black bias offset and a linear scaling coefficient that are given for each color axis of the device-dependent color space, and each color of the device-dependent color space A method that uses a common gamma for the axes.
(Appendix 23)
The method according to claim 18, wherein the individual linear correction function relating to each color in the device-dependent color space has a maximum value when the coordinate value in the device-dependent color space is a boundary value, and the device-dependent color space. A method in which the inner coordinate value decreases linearly as it approaches other boundary values or neutral axes and is a function that approaches zero or nearly zero.
(Appendix 24)
The method according to claim 19, further comprising the step of calculating a second parameter representing chromaticity among the color profiles based on the coordinate values in the linearly scaled device-independent color space, and the color profile of the input device A method of using this second parameter for obtaining
(Appendix 25)
A method for obtaining a color profile of a digital image input device, comprising:
Capturing a color sample image including a plurality of color patches with the input device;
Examining the coordinate values in the device-dependent color space of each color patch;
Estimating the illuminance of the brightest gray patch among the color patches based on the investigated device-dependent color space coordinate values;
Linearly scaling the coordinate values in the device-independent color space of each color patch based on the obtained estimated illuminance;
Calculating a first parameter representing a gradation curve of the input device based on the linearly scaled device-independent color space coordinate values;
Calculating a second parameter representing chromaticity in the color profile based on the linearly scaled device-independent color space coordinate values;
Adjusting the linearly scaled device-independent color space coordinate value or the amount derived therefrom using a separate linear correction function for each color;
The coordinate values in the device-independent color space, the first parameter, and the second parameter that have undergone the linear scaling and color-specific adjustment without using the relationship between the color sample partial exposure and the non-color sample partial exposure in the image. Obtaining a color profile of the input device based on:
Storing the obtained color profile in a computer memory;
Having a method.
(Appendix 26)
As an instruction to the programmable processor,
Instructions for receiving a color sample image including a plurality of color patches from a digital image input device;
A command to check the coordinate value in the device-dependent color space of each color patch;
A command for estimating the illuminance of the brightest gray patch among the color patches based on the coordinate values in the device-dependent color space that was examined,
A command for linearly scaling the coordinate value in the device-independent color space of each color patch based on the obtained estimated illuminance;
A command to determine the color profile of the input device based on the linearly scaled device-independent color space coordinate values;
Computer memory for storing.

It is a figure which shows the conventional image input / output system. 1 is a diagram illustrating a digital image input device profiling system according to an embodiment of the present invention. FIG. It is a figure which shows the digital image input device profiling procedure in this embodiment. FIG. 4 is a diagram showing a procedure executed in step 314 in FIG. 3 to generate a parameter list used for final calculation of coordinate values in the device-independent color space of the color profile. FIG. 5 is a diagram illustrating a procedure executed in step 406 in FIG. 4 or independently for color-specific adjustment in a device-independent color space.

Explanation of symbols

101 subject, 102, 202 digital image input device, 103, 303, 501, 507 device-dependent RGB data, 104, 106, 317 color profile, 105, 301, 502, 505 device-independent XYZ data, 107 device Dependent CMYK data, 108 Digital image output device, 109 Subject mapping, 200 Input device profiling system, 204 Computer system, 206 Data storage system, 208 Operator, 210 Other device / computer, 300 Input device profiling procedure, 302, 304 , 306, 308, 310, 312, 314, 316, 402, 404, 406, 408, 506, 508, 510, 512 steps, 305 two-dimensional data array RGB-XYZ 305, 307 Gray patch information, 309 Brightest patch information, 311 Illuminance correction coefficient α _11LLCORR , 313 Two-dimensional data array RGB-XYZ305 _ILLCORR , 313a Two-dimensional data array RGB- (XYZ _ILLCORR ) ', 315 Parameter list, 403, 410 Tone curve Description parameter α _TC , 405, 409 Chromaticity description parameter α _C , 407, 411 Color-specific adjustment amount description parameter α _SA , 500 Device-independent color space coordinate value adjustment procedure, 503 Other parameters, 504 Color-specific adjustment amount ΔXYZ _i , 509 Adjustment coefficient for each color β _i , 511 Common adjustment amount ΔXYZ _{TOTAL for} each color.

Claims (3)

A method for obtaining a color profile of a digital image input device, comprising:
Capturing a color sample image including a plurality of color patches with the input device;
Examining the coordinate values in the device-dependent color space of each color patch;
Among the gray patches whose coordinate values in the device-dependent color space that have been examined are within a predetermined limit range, the gray patch having the maximum illuminance represented by the coordinate values in the device-independent color space is identified and identified. Estimating the illuminance of the identified gray patch based on the device-dependent color space coordinate value of the gray patch;
The value obtained by multiplying the ratio of the obtained estimated illuminance and the illuminance represented by the coordinate value in the device-independent color space of the specified gray patch by the coordinate value in the device-independent color space of each color patch is a step of Ru determined as a linear scaling already device-independent color space coordinate values,
Determining a color profile of the input device based on the linearly scaled device-independent color space coordinate value or a derived amount thereof;
Storing the obtained color profile in a computer memory;
Having a method. A method for obtaining a color profile of a digital image input device, comprising:
Capturing a color sample image including a plurality of color patches with the input device;
Examining the coordinate values in the device-dependent color space of each color patch;
Among the gray patches whose coordinate values in the device-dependent color space that have been examined are within a predetermined limit range, the gray patch having the maximum illuminance represented by the coordinate values in the device-independent color space is identified and identified. Estimating the illuminance of the identified gray patch based on the device-dependent color space coordinate value of the gray patch;
The value obtained by multiplying the ratio of the obtained estimated illuminance and the illuminance represented by the coordinate value in the device-independent color space of the specified gray patch by the coordinate value in the device-independent color space of each color patch is a step of Ru determined as a linear scaling already device-independent color space coordinate values,
Calculating a first parameter representing a gradation curve of the input device based on the linearly scaled device-independent color space coordinate values;
Calculating a second parameter representing chromaticity in the color profile based on the linearly scaled device-independent color space coordinate values;
Adjusting the linearly scaled device-independent color space coordinate value or the amount derived therefrom using a separate linear correction function for each color;
Obtaining a color profile of the input device based on the coordinate values in the device-independent color space that have undergone the linear scaling and color-specific adjustment, the first parameter, and the second parameter;
Storing the obtained color profile in a computer memory;
Having a method. As an instruction to the programmable processor,
Instructions for receiving a color sample image including a plurality of color patches from a digital image input device;
A command to check the coordinate value in the device-dependent color space of each color patch;
Among the gray patches whose coordinate values in the device-dependent color space that have been examined are within a predetermined limit range, the gray patch having the maximum illuminance represented by the coordinate values in the device-independent color space is identified and identified. An instruction for estimating the illuminance of the identified gray patch based on the coordinate value in the device-dependent color space of the gray patch;
The value obtained by multiplying the ratio of the obtained estimated illuminance and the illuminance represented by the coordinate value in the device-independent color space of the specified gray patch by the coordinate value in the device-independent color space of each color patch is A command to obtain a linearly scaled device-independent color space coordinate value ;
A command to determine the color profile of the input device based on the linearly scaled device-independent color space coordinate values;
Computer memory for storing.

Images