JP2008530952A - Color correction techniques for color profiles - Google Patents

Abstract

  Selective tristimulus correction is applied to the device independent coordinates using a piecewise linear correction function. A piecewise linear correction function produces the maximum value of the piecewise linear correction function in the corresponding device-dependent color space boundary condition and the value in its corresponding device-dependent color space on either a different boundary condition or neutral axis. Is defined such that the piecewise linear correction function linearly decreases to zero or approximately zero as. By linearly reducing the piecewise linear correction function to zero or approximately zero, the correction for one color area is smoothly reduced and mixed with the other color areas, effectively introducing artifacts or image corruption due to the correction. To be prevented.

Description

  The present invention allows selective adjustment of one or more color regions without substantial risk of subsequent artifacts or unwanted alterations to the subsequently generated image. In an aspect, it relates to correcting device independent color data in a color management profile. The present invention is particularly useful in improving the color profile of digital image input devices such as digital cameras or scanners.

  A digital image input device such as a digital camera or a scanner converts light reflected from an object into digital data representing an image of the object. Typically, digital data is divided into units, each unit describing the portion of the image or the color of a pixel. Thus, an image can be described as a two-dimensional array of pixels (x, y). In addition, each unit of digital data typically describes the color of the pixel by describing the amount, or intensity, of the primary colors of red, green, and blue that are present in the pixel. For example, the digital data may represent that the pixel at x = 0, y = 0 has a red intensity of 200, a green intensity of 134, and a blue intensity of 100 of 100, where the intensity of each primary color is represented by 8 bits. Is done. (8 bits allow for 256 combinations so each primary color can have a value between 0 and 255, in this example 255 indicates the highest intensity level, zero is no intensity, or black. Show.)

  In general, since different digital image input devices generate different digital data for the same image obtained under the same conditions, the digital data generated by the digital image input device is referred to herein as “device-dependent data”. For example, the first digital camera shows that the first pixel of the image has 200 red components, while the same pixel of the equivalent image taken under the same conditions is 202 in the second digital camera. May have red component. In another example, the first digital camera records red in the apple as 200, and the second digital camera records red in the same part of the same apple (captured under the same conditions) as 202. There are things to do. Since the device dependent data generated by a digital image input device is typically specified by the red, green, and blue color components associated with each pixel, this data is often referred to as “RGB” data.

  Differences between device dependent data from two different devices result from minor differences in the imaging component of each device. These differences cause problems when those images are output by a digital image output device such as a color inkjet printer, CRT monitor, or LCD monitor. For example, an apple image captured by the first digital camera and an apple image captured by the second digital camera appear differently when output to the same color inkjet printer.

  To further complicate matters, digital image output devices also have the same type of discrepancy between each other as differences between digital image input devices. For example, a user may want to see a red square image on one CRT monitor, while a customer may see the same image on another CRT monitor. Now assume that the digital image input device used to image the recorded red square has recorded all pixels of that red square as red = 200, green = 0, and blue = 0. In general, when the two monitors display the same image, each monitor displays a slightly different red color even when it receives the same digital data from the input device.

  The same difference generally exists when printing the same image on two different printers. It should be noted here, however, that digital image data processed by a printer typically represents each pixel in the image for the secondary colors cyan, magenta, and yellow present in that pixel. Each is described according to the amount or intensity of black. Thus, device-dependent digital image data processed by the printer is referred to as “CMYK” data (as opposed to RGB device-dependent data associated with the digital image input device) (as opposed to the monitor displaying the data in RGB format). To do).

  The color profile provides a solution to the color mismatch between the devices described above. Each digital image input device typically has its own color profile, which maps device dependent data to device independent data. Correspondingly, each digital image output device usually has its own color profile, and the device-independent data can be used by the output device to print the color represented by the device-independent data. Convert to data. The device independent data describes the color of the pixels in the image in a universal manner, i.e. in a device independent color space. The device independent color space assigns a unique value to each color, but the unique value for each color is determined using calibrated instrumentation and illumination conditions. Examples of device independent color spaces are CIEXYZ, CIELAB, CIE Yxy, and CIE LCH, which are known in the art. Device-independent data is often referred to herein as “device-independent coordinates”. Here, the device-independent data in the CIEXYZ color space is referred to as “XYZ data” or simply “XYZ”. Also, device independence in the CIELAB color space is referred to as “LAB data”, “CIELAB”, or simply “LAB”.

  In theory, a color profile allows a user to obtain an image of an object using any digital image input device and provides an accurate representation of the object image from a digital image output device. Enable to output. For example, referring to FIG. 1, an image of an object 101 is acquired using a digital image input device 102. This image is represented by RGB 103 in FIG. The image RGB 103 is then converted to device independent data (eg, XYZ 105) using the color profile 104 of the digital image input device. The device-independent data XYZ 105 is subsequently converted into device-dependent data (eg, CMYK 107) specific to the output device 108 using the output device color profile 106. The output device 108 uses this device-dependent data CMYK 107 to generate an accurate representation 109 of the object 101.

  The usefulness of color profiles is limited by how accurately they convert device-dependent data to device-independent data and vice versa. Currently, there is no way to generate a color profile that fully translates device-dependent data into device-independent data, or vice versa. Errors in the color profile become apparent when comparing a displayed image, such as an image displayed on a CRT or LCD monitor, with a hardcopy printout of this image.

  As a conventional method, there is a method for improving a color profile by correcting device-independent data associated with device-dependent data. However, these methods improve the color profile while leaving a small but significant 2-3 Delta E color error in a particular region of the color space. (The unit of 1 Delta E is known in the art and refers to one unit of Euclidean distance in the CIELAB color space.) In other words, the translation or correction performed by the conventional method is still the corrected color gamut Color gamut) introduces irregularities and is not truly linear in all areas of the color space, especially for large corrections.

  Therefore, there is a need in the art for a method for correcting a color profile or device independent color space with reduced errors.

  The above-mentioned problems in this field involve the selective adjustment of one or more color gamuts with substantially no risk of introducing artifacts or undesired alterations to subsequently generated images. Handled by a system and method for correcting an enabling color profile or device-independent color space, a technical solution is achieved. According to an embodiment of the present invention, selective tristimulus correction for device independent coordinates is performed using a piecewise linear correction function. A piecewise linear correction function yields the maximum value of the piecewise linear correction function at the corresponding device-dependent color space boundary condition, and within its corresponding device-dependent color space either at a different boundary condition or neutral axis. The piecewise linear correction function is defined to linearly decrease to zero or approximately zero as the value approaches. By linearly reducing the piecewise linear correction function to zero or approximately zero, the correction for one color region is smoothly reduced and mixed with the other color regions, thereby introducing artifacts or image corruption due to the correction. Is substantially prevented.

  The present invention will be understood from the following detailed description of the preferred embodiments, taken in conjunction with the accompanying drawings. The accompanying drawings are intended to illustrate the concept of the invention and are not necessarily to scale.

  The present invention allows one or more selected color regions in a device-independent color space to have a correction magnitude that increases as the color in the color space moves away from the selected region being corrected. Correct to reduce. Therefore, adjustments to color regions in the device-independent color space are made so that the adjustments do not pose a substantial risk of introducing artifacts or undesirable alterations to subsequently generated images. be able to. The present invention has been found useful for correcting both small color mismatches, i.e. 2-3 Delta E levels, or large color mismatch, i.e. 20-30 Delta E levels.

  FIG. 2 illustrates a system 200 for generating a color profile for a digital image input device according to an embodiment of the present invention. The system 200 can include a digital image input device 202 communicatively connected to a computer system 204, which is communicatively connected to a data storage system 206. The computer system 204 may include one or more computers connected in communication and may or may not require the assistance of an operator 208. The digital image input device 202 obtains a test image of the color chart and optionally a scene outside the color chart. Examples of the digital image input device 202 include a digital camera or a scanner. The digital image input device 202 transmits a test image to the computer system 204, which is based on the estimated illumination of the test image and color chart, as will be described in detail below with reference to FIGS. Generate a color profile for the digital image input device 202. This color profile may be stored in the data storage system 206 and output to any other device 210 or other output from the computer system 204. It should be noted that information such as information relating to the test image and color chart required by the computer system 204 can be provided to the computer system 204 by any means, and is not necessarily dependent on the digital image input device 202. .

  Data storage system 206 may include one or more computer-accessible memory. The data storage system 206 may be a distributed data storage system that includes a plurality of computer-accessible memories that are communicatively connected via a plurality of computers, devices, or both. On the other hand, the data storage system 206 need not be a distributed data storage system and, as a result, may include one or more computer-accessible memory provided within a single computer or device.

  The term “computer” refers to any data processing device such as a desktop computer, laptop computer, mainframe computer, personal digital assistant, Blackberry, and / or data processing, and / or Whether for data management and / or data handling, other devices, whether electrical, and / or magnetic, and / or optical, and / or biological components, and / or other It does n’t matter.

  The term “computer-accessible memory” includes any data storage device accessible by a computer, including volatile or non-volatile, electronic, and / or magnetic, and / or optical, and Whether or not, and including but not limited to, floppy disk, hard disk, compact disk, DVD, flash memory, ROM, and RAM.

  The term “communicatively connected” refers to between devices and / or computers and / or programs that may communicate data, whether wired, wireless, or both It is intended to include any type of connection in Furthermore, the term “communicatively connected” includes connections between devices and / or programs within a single computer, connections between devices and / or programs provided in different computers, and within computers. Includes connections between devices that are not provided at all. In this regard, although the data storage system 206 is shown separated from the computer system 204, those skilled in the art will recognize that the data storage system 206 can be fully or partially contained within the computer system 204. It will be.

  FIG. 3 illustrates a method 300 for creating a color profile for a digital image input device 202 according to an embodiment of the present invention. The method 300 is performed by the computer system 204 as hardware, software, and / or firmware in embodiments of the invention.

  In step 302, the computer system 204 receives device dependent data from the digital image input device 202 or from some other source. The device-dependent data received in step 302 is RGB data in the embodiment of the present invention. This device-dependent data represents at least a color chart test image. Here, although this application is described with respect to the use of a color chart, any object may be used that has a known device-independent value associated with that color. It should be noted that the device-dependent data received in step 302 need not include data belonging to a scene outside the color chart in the test image.

  An operator 208 who can also operate the digital image input device 202 can specify the position and orientation of the color chart in the test image. Alternatively, the orientation of the color chart may be automatically determined by the computer system 204. The average of the device dependent RGB values received at step 302 can be determined for each color patch in the color chart so that a single device dependent RGB value for each color patch is output from step 302. The device-dependent RGB values each associated with a color patch are collectively referred to as an array of device-dependent RGB values represented by RGB303.

In step 304, an array of RGB values, that is, RGB 303 is associated with device-independent data XYZ 301 measured in advance for the color chart used in the test image. The association between the device-dependent data RGB 303 and the corresponding device-independent data XYZ 301 can be stored as a two-dimensional array represented as “RGB-XYZ” 305 in FIG. This array can include two columns, one for device-dependent data and one for device-independent data, where each row is associated with a single color patch in the color chart. Contains data. For example, the following software code can be used to define such a sequence.
RGB rgbChartValue [i]
Lab labMeasuredValue [i]
Here, 0 <i <numPatches.
rgbChartValue represents RGB303 in FIG. The labMeasuredValue is device-independent data XYZ301 associated with the color chart, measured in advance, converted, and stored as Lab data. numPatches represents the number of patches in this color chart. As those skilled in the art will recognize, any manner of expressing the relationship between device-dependent data RGB 303 and device-independent data XYZ 301 can be used.

  Device-independent data XYZ 301 is illustrated as being in the CIEXYZ device-independent color space, but any device-independent color space can be used. Further, FIGS. 3, 4, and 5 illustrate device-independent data as XYZ data, but individual steps in FIGS. 3, 4, and 5 can be performed on device-independent data in other color spaces. It can also act against. Therefore, conversion of XYZ data to other color spaces is implied in these steps and is not shown because such conversion is known to those skilled in the art.

ｌａｂＭｅａｓｕｒｅｄＶａｌｕｅ［ｉ］．ａ（）は、現在のパッチｉの「ａ」値（ＬＡＢ内）である。同様にｌａｂＭｅａｓｕｒｅｄＶａｌｕｅ［ｉ］．ｂ（）は、現在のパッチｉの「ｂ」値（ＬＡＢ内）である。ＧｒａｙＲａｎｇｅは、色がグレイであるか否かの決定に使用される値である。本発明の実施態様によれば、ＧｒａｙＲａｎｇｅが５デルタＥになる。
ＬａｂＧｒａｙＶａｌｕｅ．ｌａｂ［］は、出力時にＣＩＥＬＡＢデバイス非依存色空間内におけるグレイパッチ３０７の配列を含む。
ｒｇｂＧｒａｙＶａｌｕｅｓ［］は、カラーチャートのＲＧＢイメージから抽出されるデバイス依存グレイ値を表す。
ｎｕｍＧｒａｙＰａｔｃｈｅｓは、識別されたグレイパッチの総数である。 In step 306, gray patches in device independent data XYZ 301 are identified. Gray patches can be identified using device independent data XYZ stored in array RGB-XYZ (array RGB-XYZ) 305. The gray patch output identified in step 306 is represented as GrayPatches 307. The RGB-XYZ 305 can be passed to the step 308 via the step 306 without being changed. Although not required, it may be advantageous to identify gray patches in the LAB color space as opposed to the XYZ color space. In particular, in the LAB color space, the gray color is associated with coordinates “a” and “b” that are approximately equal to zero, thereby providing a simple calculation. Therefore, the device-independent data XYZ in the array RGB-XYZ 305 can be converted to LAB prior to execution of step 306. The following software code illustrates the advantages of operating in the LAB color space, including how to identify GrayPatches 307.

labMeasuredValue [i]. a () is the “a” value (in LAB) of the current patch i. Similarly, labMeasuredValue [i]. b () is the “b” value (in LAB) of the current patch i. GrayRange is a value used to determine whether the color is gray. According to an embodiment of the present invention, GrayRange is 5 Delta E.
LabGrayValue. lab [] includes an array of gray patches 307 in the CIELAB device independent color space when output.
rgbGrayValues [] represents a device-dependent gray value extracted from the RGB image of the color chart.
numGrayPatches is the total number of identified gray patches.

  In accordance with an embodiment of the present invention, Gray Patches 307 performs a device-independent measured data array (LabGrayValue [] in the software code above) as well as the corresponding device-dependent gray values ( Including both rgbGrayValues []) in the above software code. Unless it is known in advance that the chart data has already been sorted in that manner, it may be advantageous to sort the arrays of LabGrayValue [i] and rgbGrayValues [i], where the index i = 0 ,. . . , NumGrayPatches map to brightest-> darkest.

ｍａｘＧｒａｙＰａｔｃｈＩｎｄｅｘは、上記のコードの完了時に、もっとも明るい範囲内グレイパッチをストアしている配列ｌａｂＧｒａｙＶａｌｕｅ．ｌａｂ［］（ＧｒａｙＰａｔｃｈｅｓ３０７）内の位置である。ｍａｘＬは、上記のコードの完了時におけるｌａｂＧｒａｙＶａｌｕｅ．ｌａｂ［］（ＧｒａｙＰａｔｃｈｅｓ３０７）からのもっとも明るい範囲内グレイパッチのＬ＊値である。ｍａｘＬは、続いてＹ_measuredに変換されるが、それについては次に述べる。ｒｂｇＣｈａｒｔＶａｌｅｓ［］は、ｒｇｂＣｈａｒｔＶａｌｕｅｓ［ｉ］．ｒ（）がレッド成分となり、ｒｇｂＣｈａｒｔＶａｌｕｅｓ［ｉ］．ｇ（）がグリーン成分となり、ｒｇｂＣｈａｒｔＶａｌｕｅｓ［ｉ］．ｂ（）がブルー成分となるように、ＲＧＢ−ＸＹＺ３０５からのデバイス依存データを表す。 In step 308, the brightest in-range gray patch in GrayPatches 307 is identified and output as BrighttestPatch 309. According to an embodiment of the invention, the brightest in-range gray patch is a gray patch (in device independent coordinates) with associated device dependent data of R <255, G <255, and B <255. In other words, the brightest gray patch with device-dependent data that is not associated with overexposure is identified. The device-independent data and device-dependent data necessary to execute step 308 can be obtained from the array RGB-XYZ 305. Gray Patches 307 and RGB-XYZ 305 can be passed through step 308 to step 310 without modification. The following software code is sorted in the order of the intensity (coordinate “L” in the LAB) (from lightest to darkest) in order to perform the step 308, the array RGB-XYZ305 including the input GrayPatches 307. It is illustrated on the assumption that it is.

maxGrayPatchIndex, upon completion of the above code, contains an array labGrayValue. It is a position in lab [] (Gray Patches 307). maxL is the value of labGrayValue. L * value of the brightest in-range gray patch from lab [] (GrayPatches 307). maxL is subsequently converted to Y _measured , which is described next. rggChartVales [] is rgbChartValues [i]. r () becomes a red component, and rgbChartValues [i]. g () becomes a green component, and rgbChartValues [i]. The device-dependent data from RGB-XYZ 305 is represented so that b () becomes a blue component.

In steps 310 and 312, the device-independent data of the color chart (stored in RGB-XYZ 305) is scaled by the illuminance correction coefficient α _ILLCORR to compensate for the effects of excessive or insufficient illumination on the color chart. Is done. α _ILLCORR is _generated according to an embodiment of the present invention by comparing the illuminance of _{BrighttestPatch 309} in device-independent coordinates with the illuminance of the same patch indicated by the associated device-dependent coordinates. In other words, the measured illuminance of BrighttestPatch 309 is compared against the illuminance of the same patch as recorded in RGB data 303 by the digital image input device. According to an embodiment of the present invention, α _ILLCORR is generated as follows.
α _ILLCORR = (Y _estimated ) / (Y _measured )
Y _estimated is the evaluated illuminance of RGB 303 for the brightest in-range gray patch. Y _estimated can be expressed as follows.
_{Y estimated = f Y (Ri 0} , Gi 0, Bi o)
Here, the index “i ₀ ” refers to maxGrayPatchIndex. Y _measured is the illuminance of XYZ 301 as identified by BrighttestPatch 309.

右辺のベクトルはＸＹＺデータ３０１を表し、左辺のベクトルはＸＹＺ_ILLCORRを表す。この補正は、もっとも明るい範囲内グレイパッチが、２．２の初期カメラガンマを前提とするそのパッチのためのオリジナルのＲＧＢ値に対応する値を引き続き有することを保証する。本発明のこの実施態様によれば、ガンマの値およびＲＧＢ色度が、デフォルトのデジタルイメージ入力デバイスのプロファイル設定、たとえばこの分野で公知のｓＲＧＢまたはアドビ（Ａｄｏｂｅ）ＲＧＢに初期化されていることが仮定される。 In step 312, in an embodiment of the present invention, illumination corrected device independent data is generated by linear scaling of device independent data XYZ301 by α _ILLCORR . Such illuminance-corrected device-independent data is represented as XYZ _{ILLCORR in} FIG. A one-to-one color patch relationship between XYZ _ILLCORR and RGB data 303 is represented as RGB-XYZ _ILLCORR 313. In an embodiment of the present invention, XYZ _ILLCORR is generated by multiplying all values of XYZ data 301 by α _ILLCORR as illustrated below.

The vector on the right side represents XYZ data 301, and the vector on the left side represents XYZ _ILLCORR . This correction ensures that the brightest in-range gray patch will continue to have a value corresponding to the original RGB value for that patch assuming an initial camera gamma of 2.2. According to this embodiment of the present invention, the gamma value and RGB chromaticity are initialized to a default digital image input device profile setting, for example sRGB or Adobe RGB known in the art. Assumed.

  In step 314, a list of optimized parameters 315 used to generate the final device independent coordinates of the profile of the digital image input device is calculated. FIG. 4, described next, illustrates an inventive method for generating such parameters according to an embodiment of the present invention. However, any method known in the art for calculating these parameters can be used. Starting with this list of parameters 315, RGB 303 and XYZ 301 are used in step 316 to generate a profile 317 for the digital image input device 202.

FIG. 4 illustrates an expanded view of the substeps that can be performed in step 314 according to an embodiment of the present invention. The input to step 402 is an RGB-XYZ _ILLCORR array 313 of associated device dependent data RGB and illumination corrected device independent data XYZ _ILLCORR . In step 402, the optimum parameter α _TC 403 describing the one-dimensional gradation curves R, G, and B of the device-dependent data RGB (in RGB-XYZ _ILLCORR ) is generated. RGB-XYZ _ILLCORR 313 is passed through step 402 to step 404 without being changed.

インデックス「ｉ」は、カラーチャート上の「ｎ」個のグレイパッチを参照しており、インデックス「ｉ₀」は、もっとも明るい範囲内グレイパッチをストアしている配列ｌａｂＧｒａｙＶａｌｕｅ．ｌａｂ［］（ＧｒａｙＰａｔｃｈｅｓ３０７）内の場所であるｍａｘＧｒａｙＰａｔｃｈＩｎｄｅｘを参照している。（Ｌ＊ａ＊ｂ＊_ILLCORR）ベクトルは、ＸＹＺ_ILLCORRに対応するＬＡＢ座標値を表す。（Ｆ_LAB（ＲＧＢ_i，α_TC））ベクトルは、デバイス依存データＲＧＢ３０３に対応する予測されるデバイス非依存座標の計算に使用される関数を表す。この分野で公知の任意の関数をＦ_LAB（ＲＧＢ_i，α_TC）のために使用できる。ニュートン法またはパウエル法等の公知の方法を使用し、パラメータα_TCを、最小二乗誤差関数Ｅｒｒ（α_TC）の値を最小にするような方法で自動的に変更することができる。結果として得られるα_TCの値は、いくつかの場合に、ＲＧＢ３０３からのデバイス非依存座標の予測に使用するための最適値となり得る。 Process optimized alpha _TC 403, according to an embodiment of the present invention, selects an initial parameter for the alpha _TC, it applies the first to the selected parameters for the alpha _TC to the device-dependent data RGB This can be done by generating device-independent coordinate values that are predicted. Thereafter, an error between the generated predicted device-independent coordinate value and the measured device-independent coordinate value (XYZ _ILLCORR ) is calculated. This process is repeated with the newly selected parameters for α _TC until the calculated error is minimized. The parameter for α _TC 403 from which the minimum error was obtained is output as the optimum parameter α _TC 403 in step 402. The process of step 402 can be expressed as follows according to an embodiment of the present invention.

The index “i” refers to “n” gray patches on the color chart, and the index “i ₀ ” is an array labGrayValue. Reference is made to maxGrayPatchIndex which is a place in lab [] (GrayPatches 307). The (L * a * b * _ILLCORR ) vector represents the LAB coordinate value corresponding to XYZ _ILLCORR . The (F _LAB (RGB _i , α _TC )) vector represents a function used to calculate the predicted device-independent coordinates corresponding to the device-dependent data RGB303. Any function known in the art can be used for F _LAB (RGB _i , α _TC ). A known method such as the Newton method or the Powell method can be used to automatically change the parameter α _TC in such a way as to minimize the value of the least square error function Err (α _TC ). The resulting α _TC value may be an optimal value for use in predicting device-independent coordinates from RGB 303 in some cases.

ｆ_Rは、デジタルイメージ入力デバイスがデジタルカメラの場合に、Ｒ_Max、Ｒ_bias、γ_R、およびβ_cがα_TCを表すようにレッド階調チャンネルの１次元応答を定義する関数である。上記の式は、グリーンおよびブルーの階調チャンネルについても同様に使用でき、そこにおいて次式によって定義されるコントラスト関数ｆ_c（Ｒ）内で、β_c＝０である。
ｆ_c（ｘ，β_c）＝ｘ＋β_c（０．５＋０．５（２（ｘ−０．５））³−ｘ）
ここで注意を要するが、β_c＝０についてコントラスト関数ｆ_c（ｘ）＝ｘである。Ｒ_Maxは線形スケーリング係数であり、デジタルイメージ入力デバイスによって記録される最大のレッド色を示す。Ｒ_biasはブラックバイアスオフセットであり、デジタルイメージ入力デバイス２０２によって記録されるもっとも暗いレッド色を示す。γ_Rは、この分野で公知の、レッド色に関連付けされるこのデジタルイメージ入力デバイスのガンマである。それに代えてγ_Rをγ_Gおよびγ_Bと等しくし、すべてをデバイス２０２の全体的なガンマと等しくすることもできる。β_cは、コントラストを調整するためのパラメータ、すなわち値０．０＜ｘ＜０．５について関数ｆ_c（）の出力を減少し、０．５＜ｘ＜１．０について関数ｆ_c（）の出力を増加するための手段である。このような調整を実行できる多くの数学的関数が存在し、たとえば３次の多項式、スプライン等がある。そのような調整を実行するコンセプトは、この分野で公知であるが（アドビフォトショップ（Ａｄｏｂｅ（登録商標）ＰｈｏｔｏＳｈｏｐ（登録商標））等の「自動コントラスト」機能参照）、しばしばそれは、色データに関連してデバイスの特性記述を試みるというよりはイメージに対する審美的改善に関係する。 According to an embodiment of the present invention, the following equation can be used as F _LAB (R _i , α _TC ) for the red tone channel.

f _R is a function that defines the one-dimensional response of the red tone channel such that R _Max , R _bias , γ _R , and β _c represent α _TC when the digital image input device is a digital camera. The above equation can be used for green and blue tone channels as well, where β _c = 0 within the contrast function f _c (R) defined by:
f _c (x, β _c ) = x + β _c (0.5 + 0.5 (2 (x−0.5)) ³ −x)
Note that the contrast function f _c (x) = x for β _c = 0. R _Max is a linear scaling factor and indicates the maximum red color recorded by the digital image input device. R _bias is the black bias offset and indicates the darkest red color recorded by the digital image input device 202. γ _R is the gamma of this digital image input device associated with the red color, known in the art. Alternatively, γ _R can be equal to γ _G and γ _B, and all can be equal to the overall gamma of device 202. beta _c is a parameter for adjusting the contrast, that is, the value 0.0 <reducing the output of the function f _c () for x <0.5, 0.5 <x <1.0 for function f _c () Is a means for increasing the output of. There are many mathematical functions that can perform such adjustments, such as cubic polynomials, splines, and the like. The concept of performing such adjustments is well known in the art (see “Auto Contrast” functions such as Adobe® PhotoShop®), but often it relates to color data Rather than trying to characterize the device, it relates to aesthetic improvements to the image.

ｐａｒａｍｅｔｅｒＬｉｓｔはステップ４０２によって出力されるα_TC４０３を表し、ＮｕｍＰａｒａｍｅｔｅｒｓはα_TC４０３内のパラメータの数を表す。ｂｒｉｇｈｔｅｓｔＶａｌｉｄＧｒａｙは、ＸＹＺ_ILLCORR内のもっとも明るい範囲内グレイパッチ位置を表す。ｎＧｒａｙは、グレイパッチの総数を表す。ｘｙｚＰｒｅｄは、ＲＧＢ３０３内のデバイス依存値に対応する予測したデバイス非依存値である。ｅｖａｌｕａｔｅＭｏｄｅｌ（）は、予測したデバイス非依存値ｘｙｚＰｒｅｄを生成する関数であり、上記のＦ_LAB（ＲＧＢ_i，α_TC）に類似している。ｌａｂＰｒｅｄは、ｘｙｚＰｒｅｄのＬＡＢバージョンであり、関数ｍＣｏｌＭｅｔｒｉｃ−＞ＸＹＺＴｏＬａｂ（）を使用してｘｙｚＰｒｅｄはＬＡＢ空間に変換される。ｌａｂＭｅａｓは、現在のグレイパッチのＬＡＢデバイス非依存値（ｇｒａｙＬａｂ［］）である。ｘｙｚＭｅａｓは、ｌａｂＭｅａｓのＸＹＺバージョンである。ｘｙｚＭｅａｓ＊は、α_ILLCORRを表すＹ＿ＩｌｌｕｍｉｎａｔｉｏｎＣｏｒｒによって補正されたｘｙｚＭｅａｓを表す。ｅｒｒＳｑは個別の二乗誤差であり、ｓｕｍＳｑは合計の二乗誤差である。 The following software code illustrates how step 402 can be performed in accordance with an embodiment of the present invention. In this software code, it is assumed that all Gray Patches expressed in RGB-XYZ _ILLCORR 313 are sorted in the order from the brightest to the darkest.

parameterList represents the α _TC 403 output by step 402, and NumParameters represents the number of parameters in the α _TC 403. _{brighttestValidGray} represents the brightest in-range gray patch position in XYZ _ILLCORR . nGray represents the total number of gray patches. xyzPred is a predicted device-independent value corresponding to a device-dependent value in RGB 303. evaluateModel () is a function that generates a predicted device-independent value xyzPred, and is similar to F _LAB (RGB _i , α _TC ) described above. labPred is the LAB version of xyzPred, and xyzPred is converted to LAB space using the function mColMetric-> XYZToLab (). labMeas is the LAB device independent value (grayLab []) of the current gray patch. xyzMeas is an XYZ version of labMeas. xyzMeas * represents xyzMeas corrected by Y_Illumination _Corr representing α _ILLCORR . errSq is the individual square error and sumSq is the total square error.

この例においては、α_TC４０３は、ＲＧＢＭａｘ、ｇａｍｍａ、ｂｌａｃｋＢｉａｓ、およびｆｉｌｍＣｏｎｔｒａｓｔＣｏｒｒを含む。ＲＧＢＭａｘは、前述した線形スケーリング係数Ｒ_Maxをはじめ、それに相当するＧ_MaxおよびＢ_Maxに対応する。Ｇａｍｍａは、前述したγ_Rをはじめ、それに相当するγ_Gおよびγ_Bに対応する。ｂｌａｃｋＢｉａｓは、前述したＲ_biasをはじめ、それに相当するＧ_biasおよびＢ_biasに対応する。ｆｉｌｍＣｏｎｔｒａｓｔＣｏｒｒはβ_cに対応する。ここで注意が必要であるが、ｒｇｂＩｎｄｅｘは、いずれのチャンネルが計算中であるかを示し（ｒｇｂＩｎｄｅｘ＝０，１，２＝＞Ｒ，Ｇ，Ｂ）、ｃＶａｌは、ｅｖａｌＧａｍｍａＭｏｄｅｌ（）関数によって修正された対応する出力色値に変換中の入力色値（Ｒ、Ｇ、またはＢ）である。値「ｒｅｓｕｌｔ」は、入力インデックスｒｇｂＩｎｄｅｘ＝０，１，２についてＲ，Ｇ，Ｂ（ｏｕｔ）を獲得するために計算される。ｒｅｓｕｌｔｉｎｇＲＧＢベクトルは、後述するＲＧＢ−＞ＸＹＺ行列によって乗じられ、計算後のＸＹＺの値を得る。上記の合計の二乗誤差の計算が、ＲＧＢ−＞ＸＹＺの計算においてパラメータα_TC４０３を系統的に調整した後、その合計の二乗誤差が最小化されるまで反復される。 The value of α _TC 403 needs to be adjusted to minimize the calculated sum of squared errors (eg, sumSq). The mathematical model used to predict device-independent coordinates associated with RGB (in RGB-XYZ _ILLCORR ) can be any model known in the art (eg, evaluateModel ()). However, the selected model provides smoothness (ie, low second derivative values and / or no undesirable visual banding or artifacts when this profile is used to characterize the image) However, it is important to achieve an accurate prediction with as few parameters as possible. In the embodiment of the present invention, the following software code is a mathematical model for predicting device-independent coordinates for device-dependent data RGB (evalGammaModel () used as F _LAB (RGB _i , α _TC ) described above). Illustrated.

In this example, α _TC 403 includes RGBMax, gamma, blackBias, and filmContrastCorr. RGBMax corresponds to G _Max and B _Max corresponding to the linear scaling coefficient R _Max described above. Gamma, including the gamma _R described above, corresponding to gamma _G and gamma _B corresponding thereto. blackBias corresponds to G _bias and B _bias corresponding to R _bias described above. filmContrastCorr corresponds to the β _c. Note that rgbIndex indicates which channel is being calculated (rgbIndex = 0, 1, 2> R, G, B), and cVal is modified by the evalGammaModel () function. The input color value (R, G, or B) being converted into the corresponding output color value. The value “result” is calculated to obtain R, G, B (out) for the input index rgbIndex = 0,1,2. The resulting RGB vector is multiplied by a later-described RGB-> XYZ matrix to obtain a value of XYZ after calculation. The above total square error calculation is repeated until the total square error is minimized after systematically adjusting the parameter α _TC 403 in the RGB → XYZ calculation.

Note that the above tone curve calculation starts with the contrast adjustment applied to the RGB input values. This contrast adjustment is a third order with the constraint that no correction is applied to the values RGB = 0, 0.50, or 1.0 (assuming that RGB is normalized to 1.0). It is a polynomial. This correction is multiplied by filmContrastCorr, but default = 0.0, which can be added to RGB values as positive or negative (decrease or increase contrast respectively). The corrected values are the black bias offset (defined by blackBias []), the linear scaling factor (RGBMax []) for each R, G, B channel, and the gamma value (gamma) for each RGB channel. []) Together with the standard gamma function. In this example, the total number of adjustable parameters α _TC 403 is 10, ie, RGBMax, gamma for R, G, and B, blackBias for R, G, and B, 3 With a single value filmContrastCorr for all channels. Note that if the number of data values for the color chart is a 6 gray patch, the data points to be calculated are 18 (6 × 3 (for L *, a *, b *) = 18). It becomes. If the first two patches are overexposed, the number of data points is reduced to 4 × 3 = 12. Optionally, the number of parameters α _TC 403 can be reduced to 8 assuming that the gamma is exactly the same for each of the three RGB channels. This approach is useful for lower quality images with noisy RGB data extracted from a color chart.

Once the optimal parameter α _TC 403 is determined in step 402, step 404 generates an optimal parameter α _C 405 that describes the RGB chromaticity of the device profile. α _TC 403 and RGB-XYZ _ILLCORR 313 are passed through step 404 to step 406 without being changed. In step 404, R, G, and B for all color chart values using the tone response calculated in step 402, using standard matrix / TRC format, as shown in the following description. XYZ values for can be calculated. The error function is similar to that described above in connection with step 402. However, all data points can be used and the R, G, or B value is equal to 0 or 255, and the image data is clipped and the actual R, G, or B value is not known exactly , All square errors added to the sum can be reduced by a predefined weighting factor.

Ｍ行列は、ＲＧＢチャンネルの（ｘ，ｙ）色度およびシステムのホワイトポイント、すなわちＲＧＢ＝ｍａｘにおけるＸＹＺの値から結果として生ずる（ｘ，ｙ）色度によって一意的に決定される。

この場合、誤差関数を定義するための数学的形式が、ステップ４０２に関連して前述した階調曲線を定義するプロセスと同じになるが、（ａ）すべてのカラーチャート値が使用可能であること（グレイパッチだけであったことに対して）、および（ｂ）定義されるパラメータα_Cが、Ｒ、Ｇ、Ｂチャンネルのｘおよびｙの値であることが異なる。モデルのホワイトポイント（ｘ_wp，ｙ_wp）は、チャートの測定済みデータのために使用されるホワイトポイント、通常はこの分野で公知のとおりＤ５０に対応する必要がある。 In step 404, the adjusted parameter list α _C 405 is the chromaticity values for R, G, and B (xy, known in the art). According to the embodiment of the present invention, the mathematical description of the matrix portion of the RGB XYZ function is as follows. Note that the device dependent data RGB (in RGB-XYZ _ILLCORR 313) has already been processed by the RGB tone curve function defined above in connection with step 402. In other words, the new value of RGB is in the linear RGB space. Therefore, it is possible to easily convert them to XYZ using simple matrices.

The M matrix is uniquely determined by the (x, y) chromaticity resulting from the (x, y) chromaticity of the RGB channel and the system white point, ie, the value of XYZ at RGB = max.

In this case, the mathematical form for defining the error function is the same as the process for defining the tone curve described above in connection with step 402, but (a) all color chart values are usable. (As opposed to being only a gray patch), and (b) the defined parameter α _C is the x and y values of the R, G, B channels. The model white point (x _wp , y _wp ) should correspond to the white point used for the chart's measured data, usually D50 as is known in the art.

Step 406 is optional, where optimal parameter α _SA 407 describes the selective adjustment to the device independent color space value XYZ of the resulting device profile. The parameter α _SA 407 can be optimized using the error minimization routine described above in connection with steps 402 and / or 404. According to an embodiment of the present invention, the selective adjustment performed by parameter α _SA 407 is performed on one or more colors in step 406, so that the profile modified using this approach There is virtually no risk of introducing artifacts or undesirable corruption into the converted or rendered image. Although any method can be used in step 406, an inventive process for performing selective adjustments to the device-independent color space as part of step 406 will now be described in connection with FIG. The embodiment illustrated using FIG. 5 corresponds to i = 0,... Corresponding to R, G, B, C, M, Y. . . , 5 can be used to generate a parameter α _SA 407 that contains a list of corrections ΔXYZ _i . Such parameters can be optimized using the error minimization process described above such that an adjusted device-independent color space XYZ _adj 505 is generated that minimizes the total square error.

Step 408 is also an optional step, where final optimization of some or all of the parameters in α _TC 403, α _C 405, and α _SA 407 can be performed. These parameters are individually determined by repeating steps 402 and / or 404 and / or 406 and / or using the error minimization techniques described above, such as Powell, and optimizing most of those parameters as a group. Can be optimized. For images that are not overexposed, the total number of data points is 24 × 3 = 72. The total number of adjustable parameters (α _TC 403) for the tone curve is 10, 6 for chromaticity (α _C 405) and 18 for selective XYZ adjustment (α _SA 407). Since there is a strong correlation between chromaticity values and selective XYZ adjustments, it may be advantageous to perform global optimization of the tone curve and selective chromaticity simultaneously while keeping the chromaticity fixed. The output of step 408 may include an adjusted α _C parameter α _C '409, an adjusted α _TC parameter α _TC ' 410, and an adjusted α _SA parameter α _SA '411. α _C '409, α _TC ' 410, and α _SA '411 are output from 408 and included in ParameterList 315. RGB-XYZ _ILLCORR 313 does not need to be output from step 408; instead, α _ILLCORR 311 is output from step 408 and can be added to ParameterList 315 used in step 316 to generate profile 317. In this regard, note that step 312 can be performed as part of step 316. In this scenario, the outputs α _ILLCORR 311 and RGB-XYZ ₃₀₅ from step 310 will be passed directly to step 314 as inputs instead of RGB-XYZ _ILLCORR 313. When it is necessary to operate the illuminance-corrected device-independent data in step 314, α _ILLCORR 311 can be applied to XYZ ₃₀₁ (stored in RGB-XYZ ₃₀₅₎ to generate XYZ _ILLCORR .

  The above method of FIGS. 3 and 4 reveals that it provides good results for images using color checkers known in the art that are overexposed or less illuminated than the rest of the scene. In all cases, extreme color and overall white balance are improved. If desired, modify the measured chart values XYZ 301 to obtain the equivalent of a “look profile”, ie, the image is more saturated or other modifications can be made. be able to.

  With reference now to FIG. 5, a method 500 for performing selective adjustments to a device independent color space is illustrated in accordance with an embodiment of the present invention. The method 500 ensures excellent visual alignment between the image displayed for viewing and the same image printed out in hardcopy format. The method 500 is useful for characterizing image data coming from an input device, rendering an image to a display device, and converting the image to an output device. The method 500 is also useful for improving the accuracy of color profiles for digital image input devices such as digital cameras or scanners. Accordingly, method 500 may be used as part of step 406 of FIG. However, those skilled in the art will recognize that the method 500 may be used completely independently of the method 300 illustrated in FIGS.

  Method 500 applies the correction in a linear fashion within all or selected areas of device-independent data derived from device-dependent data, with virtually no possibility of introducing artifacts. Selective adjustments are made to device-independent data: red (“R”), green (“G”), blue (“B”), cyan (“C”), magenta (“M”) Or improve the color in the selected color region, such as near yellow (“Y”), without degrading the color in the uncorrected color area.

The input used by the method 500 includes device dependent data “RGB” 501, device independent data “XYZ” 502, other parameters 503, and a change “ΔXYZ _i ” 504 to the selected color region with device independent coordinates. In which “i” represents a local color region. When method 500 is used as part of step 406 in FIG. 4, ΔXYZ _i 504 represents α _SA 407. The device dependent data RGB 501 and the device independent data XYZ 502 correspond to each other, and as a result, each element (piece) of the device dependent data in RGB 501 corresponds to the element of the device independent data in XYZ 502. For example, if method 500 is used as step 406 in FIG. 4, RGB data in RGB-XYZ _ILLCORR 313 is input as RGB data 501 and associated device-independent data XYZ _ILLCORR in RGB-XYZ _ILLCORR 313. Can be input as XYZ data 502.

When attempting to improve a profile for RGB device-dependent data, such as RGB data 501, it is desirable to perform corrections in a device-independent color space similar to the output of mathematical expressions describing device characteristics associated with RGB data. A general formula that characterizes RGB data generation devices such as digital cameras and scanners is a modified version of the matrix / TRC format described above in connection with step 402 of FIG. 4, for example. This matrix / TRC format inherently converts RGB to XYZ. Therefore, for most devices, the XYZ tristimulus space is the preferred space for performing correction. Thus, device independent data 502 is shown as XYZ, and the output of method 500 is the adjusted XYZ coordinates illustrated as XYZ _adj 505. However, linear corrections similar to those described with reference to method 500 have been used with some degree of success for visually uniform color spaces such as CIELAB known in the art, and various CIECAM models such as CIECAM98. Can be executed with.

ΔXYZ _i 504 represents a change (eg, saturation, hue, and luminance) to the selected color region “i” in the device-independent color space. According to an embodiment of the present invention, 0 ≦ i ≦ 5, and values from 0 to 5 are associated with each of red, green, blue, cyan, magenta, and yellow (R ₀ , G ₀ , B ₀ , respectively). , C ₀ , M ₀ , and Y ₀ ). However, those skilled in the art will recognize that changes to different color regions can also be used. The value R ₀ is defined as R = R _max , G = 0, B = 0, C = 0, M = 0, and Y = 0. The values of G ₀ , B ₀ , C ₀ , M ₀ , and Y ₀ follow a similar form. ΔXYZ _i 504 can be generated manually or automatically. An example of manual generation of ΔXYZ _i 504 is to have an operator observe the displayed image or hardcopy printout of an image generated from device-independent data XYZ data 502 (eg, on a CRT) and subsequently displayed or printed. Make changes to select the color of the image. For example, if red in the displayed image appears too bright, the operator can specify a negative luminance value for ΔXYZ ₀ 504 to reduce the luminance of red in the image. Alternatively, ΔXYZ _i data 504 can be automatically generated through the use of an error minimization routine, as described above in connection with steps 402 and / or 404. For example, if the method 500 is used as part of step 406 of FIG. 4, an initial value for the ΔXYZ _i data 504 can be automatically generated. Thereafter, different values of ΔXYZ _i data 504 are used in an attempt to minimize the total square error between the resulting device-independent color space XYZadj 505 and the predicted device-independent color space. Method 500 can be performed iteratively. The value of ΔXYZ _i data 504 attributed to the minimum total square error can be output as α _SA 407 at step 406.

Step 506 of method 500 takes RGB 501 and other parameters 503 as input and generates a linear RGB value (RGB) '507. Alternatively, step 506 can take a linear value (RGB) ′ directly as input without the need for other parameters 503 and step 506. Although any procedure can be used to generate (RGB) '507, (RGB)' 507 was usually multiplied by the appropriate RGB-> XYZ matrix as described in the optimization process above. Sometimes calculated to be a corrected value of RGB 501 that achieves an optimized prediction of XYZ 502. In the case of CRT, (RGB) ′ can be calculated by the form of the gamma curve associated with the CRT, eg, Rγ. In this case, the other parameters 503 are parameters describing the gamma curve. More complex functions beyond the use of mere gamma curves may be needed for higher accuracy or for LCD systems. If method 500 is used as step 406 of FIG. 4, other parameters 503 may include one or more of the parameters in α _TC 403. For example, when device-independent coordinates of a digital camera profile are adjusted, the camera's tone channel representations f _R , f _G , f _B (and associated α _TC 403 parameters as described in connection with step 402). Can be used in step 506 to generate (RGB) '507.

In step 508, a correction coefficient β509 for each color i corresponding to the ΔXYZ _i data 504 is generated. In an embodiment of the present invention, the correction factor β509 uses a piecewise linear correction function to produce a respective maximum value of the piecewise linear correction function in the corresponding device dependent color space (RGB501) boundary condition, and Each piecewise linear correction function is calculated to linearly decrease to zero or approximately zero as its corresponding value in device-dependent color space reaches either a different boundary condition or neutral axis. In other words, the correction factor β 509 is calculated according to an embodiment of the present invention, in which (a) a piecewise linear correction function for the correction factor (β) and color i can be associated with the device-dependent color currently being evaluated for i. Is maximized when at the maximum distance from all colors and neutral axes, and (b) a piecewise linear correction function for correction factor (β) and color i allows the device-dependent color under evaluation to be associated with i Is calculated to linearly reduce to zero as it approaches the color and neutral axes. To illustrate what is meant by “association”, consider the color R _b G _b B _b in the boundary condition. For example, here, the boundary condition is defined as all values of R _b G _b B _b are either 0 or 1.0. “Adjacent” or “related” colors are those in which only one of those colors has changed from 0 to 1 or from 1 to 0 from the current color. Therefore, the colors adjacent to R _b G _b B _b = (1, 0, 0) are (1, 1, 0) and (1, 0, 1). The colors adjacent to R _b G _b B _b = (1,1,0) are (1,0,0) and (0,1,0).

たとえばｉ＝０（レッドに関連付けされる）であれば、β₀は、評価中の現在のデバイス依存色（（ＲＧＢ）’５０７から）がｉに関連付けできるほかのすべての色（すなわちグリーン、ブルー、シアン、マゼンタ、およびイエロー）および中立軸から最大距離にあるときに最大になる。言い換えるとβ₀は、現在のデバイス依存色（（ＲＧＢ）’５０７から）のグリーンおよびブルー成分がゼロと評価されるときに最大になる。β₀のための区分的線形補正関数は、評価中のデバイス依存色（（ＲＧＢ）’５０７から）がグリーン、ブルー、シアン、マゼンタ、イエロー、または中立軸に近づくとゼロに線形縮小される。言い換えるとβ₀に関連付けられた線形補正関数は、評価中の現在のデバイス依存色（（ＲＧＢ）’５０７から）のグリーンおよびブルー成分が増加するに従って、ゼロに向かって線形縮小する。 In accordance with an embodiment of the present invention, β for i = 0 to i = 5 for ΔXYZ _i data 504 (ie, for R ₀ , G ₀ , B ₀ , C ₀ , M ₀ , and Y ₀ ). _i 509 is calculated as follows. According to this embodiment, it is assumed that R, G, and B are normalized to 1.0, so that β is between zero and 1, and the maximum correction occurs at β = 1.

For example, if i = 0 (associated with red), β ₀ is all other colors (ie green, blue) that the current device dependent color under evaluation (from (RGB) '507) can associate with i. , Cyan, magenta, and yellow) and the maximum distance from the neutral axis. In other words, β ₀ is maximized when the green and blue components of the current device dependent color (from (RGB) '507) are evaluated as zero. The piecewise linear correction function for β ₀ is linearly reduced to zero when the device-dependent color under evaluation (from (RGB) '507) approaches the green, blue, cyan, magenta, yellow, or neutral axis. In other words, the linear correction function associated with β ₀ linearly shrinks towards zero as the green and blue components of the current device-dependent color under evaluation (from (RGB) '507) increase.

In step 510, each individual device-independent change ΔXYZ _i 504 is corrected by a corresponding correction factor β _i 509. The total adjustment performed on the device-independent data XYZ 502 is calculated as ΔXYZ _total 511. ΔXYZ _total 511 is calculated by summing each individually corrected device independent change ΔXYZ _i 504. According to an embodiment of the present invention, ΔXYZ _total 511 is calculated as follows.

In step 512, the device-independent data XYZ 502 is adjusted by ΔXYZ _total 511, and XYZ _adj 505 is generated. According to an embodiment of the present invention, XYZ _adj 505 is calculated as follows.
XYZ _adj = XYZ + ΔXYZ _total

The following software code illustrates an implementation of the method 500 according to an embodiment of the present invention.

The method 500 can be used to improve the process of soft proofing on a display. In order to determine the desired correction to achieve such an improved soft proof, the correction magnitude and direction is adjusted to display profile A-> display profile A 'in the desired direction of improvement, and then Evaluation can be performed by converting the color of the corrected display profile A ′-> display profile A. For example, if the operator wishes to shift the yellow hue in the displayed image towards red, Δ (XYZ) ₅ (equal to a 3 delta E hue shift towards red for saturated yellow. Yellow) values can be used. When the operator confirms that the desired result has occurred, this time corrects the RGB profile of the display in the opposite manner to ensure that the image rendered on the screen has the desired red shift in yellow. It can be secured. This can be achieved with reasonable accuracy by adjusting the display profile A using the value of Δ (XYZ) ₅ (yellow), which is the negative of the above correction.

  Similarly, the method 500 can be used to perform device independent color space correction. For example, the original RGB profile (profile A1) can be modified according to method 500 to generate a new profile (profile A2). The correction of XYZ can be performed by following the standard color management path, XYZ-> RGB (profile A1)-> RGB (profile A2)-> (XYZ) '. Combining multiple transformations into a single transformation is known in the art and is called profile concatenation. The result of XYZ-> (XYZ) 'is known as an abstract profile.

  Another application for method 500 is profiling of a digital image input device (such as a scanner or digital camera) as described above in connection with FIGS. The method 500 preserves gray balance and good overall image integrity, and improves the predictive transformation of RGB-> XYZ or L * a * b * for certain objects in images with strong colors. In addition, appropriate corrections can be applied to a comprehensive RGB profile for a digital camera or scanner. The correction values for angle, saturation, and illuminance for RGBCMY can be automatically calculated by an error minimization routine.

  Unlike other techniques, the method 500 can be used in part for both small, i.e. 2-3 Delta E color mismatch, or large, 20-30 Delta E color mismatch. Use linear functions.

It is the figure which showed the conventional structure for acquiring and outputting an image. FIG. 2 illustrates a system for generating a color profile for a digital image input device according to an embodiment of the present invention. FIG. 5 illustrates a method for generating a color profile for a digital image input device according to an embodiment of the present invention. FIG. 4 illustrates a method for calculating a parameter list used to calculate a final device independent coordinate of a profile that can be used as step 314 of FIG. 3 according to an embodiment of the present invention. FIG. 6 illustrates a method for performing selective adjustments to a device-independent color space that can be used as part of step 406 of FIG.

Explanation of symbols

101 object, 102 input device, 103 device dependent data RGB, 104 color profile, 105 device independent data XYZ, 106 output device color profile, 107 device independent data CMYK, 108 output device, 109 Representation, 200 systems, 202 input devices, 204 computer systems, 206 data storage systems, 208 operators, 210 optional devices / computers, 300 methods, 301 XYZ data, 302 steps, 303 RGB data, 304 steps, 305 RGB-XYZ, 306 steps, 307 Gray Patches (gray patch), 308 steps, 309 Brighttest Patch, 310 steps, 311 α _{ILLC ORR} , 312 step, 313 RGB-XYZ _ILLCORR , 314 step, 315 parameter, 316 step, 317 profile, 402 step, 403 parameter α _TC , 404 step, 405 parameter α _C , 406 step, 407 parameter α _SA , 408 step, 409 Adjusted α _C parameter α _C ′, 410 Adjusted α _TC parameter α _TC ′, 411 Adjusted α _SA parameter α _SA ′, 500 method, 501 Device dependent data “RGB”, 502 Device independent data “XYZ”, 503 other parameters, changes to the 504 selected color region (change) "DerutaXYZ _i", 505 adjusted device-independent data XYZ _adj, 506 steps, 507 adjusted device dependent data (RGB) ', 508 steps, 509 Positive factor beta, 510 steps, 511 overall changes ΔXYZ _total, 512 steps to the device-independent data.

Claims (23)

A method executed at least in part by a computer,
The method comprises
Applying selective tristimulus correction to device independent coordinates using a piecewise linear correction function,
The piecewise linear correction function produces a maximum value of the piecewise linear correction function at a boundary condition of the corresponding device dependent color space, and within the corresponding device dependent color space at either a different boundary condition or neutral axis The piecewise linear correction function is defined to linearly decrease to zero or approximately zero as the value of. The method of claim 1, wherein
The piecewise linear correction function depends on linear device dependent coordinates associated with an RGB profile. The method of claim 1, wherein
Furthermore, the application of a combination of piecewise linear correction functions corresponding to different color regions in the device dependent coordinate space, each correction function associated with the RGB profile,
Each of the piecewise linear correction functions yields a respective maximum value of the piecewise linear correction function at each boundary condition corresponding to the device-dependent color space, and either a different respective boundary condition or the neutral axis A method in which the piecewise linear correction function is defined to linearly decrease to zero or approximately zero, respectively, as the value in the corresponding device dependent color space approaches. The method of claim 1, wherein
Further, the selective tristimulus correction is applied to the device-independent coordinates of the RGB profile that uses the piecewise linear correction function and connects the original and modified profiles to create an abstract profile. Adjusting the color space by: The method of claim 1, wherein
The method further includes modifying an RGB profile by applying the selective tristimulus correction to the device independent coordinates using the piecewise linear correction function. The method of claim 5, wherein
The RGB profile includes a digital camera or scanner profile. The method of claim 1, wherein
The method further includes adjusting the appearance of the image by applying the selective tristimulus correction to the device independent coordinates using the piecewise linear correction function.   The method of claim 1, wherein the device independent coordinates define colors in a visually non-uniform color space. A method executed at least in part by a computer,
The method
Determining tristimulus corrections for device-independent coordinates corresponding to each of the plurality of color regions in the device-dependent coordinate space based at least on the linear device-dependent coordinates;
Applying tristimulus corrections to device-independent coordinates in the corresponding color region to obtain corrected device-independent coordinates;
The tristimulus value correction is
The maximum value of the tristimulus correction occurs in the boundary condition of the corresponding device-dependent color space;
And a method in which the tristimulus correction is defined to linearly decrease to zero or approximately zero as the corresponding value in the device dependent color space approaches either a different boundary condition or neutral axis.   10. The method of claim 9, wherein the determination of tristimulus value correction further includes calculating a correction factor corresponding to each color region.   The method of claim 10, wherein the linear device-dependent coordinates include linear RGB coordinates.   The method of claim 9, wherein the color regions in the device dependent coordinate space include red, green, blue, cyan, magenta, and yellow. A method,
Determining a correction level for a chromatic color in a device-dependent color space associated with a display device based at least on a desired change in saturation, hue, and brightness;
Calculating a correction factor based at least on a linear correction function associated with the display device;
Applying the correction factor and the correction level to device-independent coordinates defining a chromatic color of the printing device to create chromatically corrected device-independent coordinates;
The correction factor produces a maximum value of the correction factor in a boundary condition of the device dependent color space, and the correction coefficient is zero or as the value in the device dependent color space approaches either a different boundary condition or a neutral axis A method defined to linearly decrease to approximately zero.   The method of claim 13, wherein calculating the correction factor further includes calculating a piecewise linear correction function of linear device dependent coordinates.   The method of claim 14, wherein the linear device dependent coordinates are linear RGB coordinates. 15. The method of claim 14, wherein
further,
Calculate a group of multiple piecewise linear correction functions, each corresponding to a different color region,
Applying each piecewise linear correction function in the group to device-independent coordinates in the corresponding color region;
Including methods. Computer accessible memory,
Instructions for causing the processor to apply selective tristimulus correction to device independent coordinates using a piecewise linear correction function;
The piecewise linear correction function is
As the maximum value of the piecewise linear correction function occurs at a corresponding device dependent color space boundary condition and the value in the corresponding device dependent color space approaches either a different boundary condition or a neutral axis, the piecewise linear correction function A computer-accessible memory defined such that the linear correction function linearly decreases to zero or approximately zero.   The computer readable medium of claim 17, wherein the piecewise linear correction function operates on linear device dependent coordinates associated with an output device. The computer-readable medium of claim 17.
Furthermore, for the processor,
Generating a group of piecewise linear correction functions, each group corresponding to a different color region in a device dependent coordinate space associated with the output device; and
Instructions for applying each group of piecewise linear correction functions to device-independent coordinates in the corresponding color region to perform tristimulus corrections on the device-independent coordinates;
Each of the piecewise linear correction functions is
Each maximum value of the piecewise linear correction function occurs at a respective boundary condition of the corresponding device-dependent color space, and within the corresponding device-dependent color space either at a different respective boundary condition or the neutral axis. A computer readable medium defined such that as the value approaches, the piecewise linear correction function decreases linearly to zero or approximately zero, respectively. The computer readable medium of claim 17.
A computer readable medium further comprising instructions that cause the processor to adjust color space by applying the selective tristimulus correction to the device independent coordinates using the piecewise linear correction function. The computer readable medium of claim 17.
A computer readable medium further comprising instructions that cause the processor to modify the RGB profile by applying the selective tristimulus correction to the device independent coordinates using the piecewise linear correction function.   The computer readable medium of claim 21, wherein the RGB profile comprises a digital camera or scanner profile. The computer readable medium of claim 17.
Further, a computer readable medium that causes the processor to adjust the appearance of an image by applying the selective tristimulus correction to the device independent coordinates using the piecewise linear correction function.

Images