Data processing: measuring – calibrating – or testing – Calibration or correction system – Zero-full scaling
Reexamination Certificate
2000-04-26
2004-07-20
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Calibration or correction system
Zero-full scaling
C382S312000, C382S155000, C358S504000, C386S349000, C430S359000
Reexamination Certificate
active
06766263
ABSTRACT:
TECHNICAL FIELD
This invention relates generally to the calibration of digital graphic capture devices and, more particularly, to a calibration procedure for digital capture devices based on the human visual system perception of colors to ensure high color fidelity of the captured image.
BACKGROUND OF THE INVENTION
Color calibration of digital image capture devices is important to maintain a high fidelity of the captured image so that a representative image of the item captured may be displayed to a consumer or user of the digital image. This high fidelity becomes more and more important as web-based commerce continues to expand. This may be appreciated as consumers begin to rely more heavily on these images and the accuracy of the colors represented thereon for their purchase decisions. For consumers to continue to expand their usage of web-based commerce, a certain minimum acceptable level of color fidelity quality must be established. Further, this color fidelity must be such that the digital counts output from the capture devices accurately represent the colors as they would appear to a typical human observer. An accurate relationship is necessary to enable reliable digital commerce where users make purchase decisions based upon the trust that the digitally captured image represents the original color appearance of the captured object that is represented in the digital image.
While various methods exist for the calibration of these digital image capture devices, these methods have, to date, required the acquisition of expensive calibration equipment (in the range of $25,000) and the employment of highly trained color scientists. Examples of these calibration procedures may be found in various books and standards, such as Color Management (GATF), Color Management (Giorgianni), the IEC 61966-9 draft specification, and the ISO 17321 draft specification. Calibration facilities and procedures may also be contracted out to specialized laboratories. However, the cost of such calibration services is typically in the range of $20,000 per calibration. As a result, many vendors of these capture devices, especially second and third tier vendors of inexpensive equipment, do not perform this calibration, or perform only a simplified calibration. As a result, many of the images published on the web or used in other media that have been captured by these devices have questionable color fidelity.
SUMMARY OF THE INVENTION
The inventive concepts of the instant invention involve the calibration of digital image capture devices such as digital cameras, scanners, etc. Recognizing the importance of high color fidelity in a competitive commercial environment which will not bear the tremendous expense of prior color calibration methods, the instant invention provides a simple and inexpensive method to ensure a high level of color fidelity that is acceptable for conducting true e-commerce. This method also reduces the error introduced by even the most expensive calibration equipment into the calibration process, and does so with inexpensive and custom color targets. It performs its calibration through the use of a simple regression, such as may be performed by a conventional electronic spreadsheet program, e.g. Excel, etc. Of course, a specialized regression algorithm could be used instead of a conventional spreadsheet, and a multi-linear regression could replace the simple regression, depending on the system requirements. As a result of this method, all digital image capture devices may deliver the high fidelity required in today's e-commerce environment.
Specifically, the method begins by simply acquiring a test target, such as an ANSI IT8, IEC, Q60, MacBeth, etc. color characterization target and its measurement values, including the white point values. This procedure also allows for the use of custom color targets as desired. Preferably, the target provides at least 6 points, although the calibration method achieves acceptable results with as little as 3 color points plus the white and black samples. Higher color fidelity may be achieved by providing more points. This step does not require that the colors be measured in any way, only that the values of the targets be provided in a physical color space, e.g. CIE 1931 XYZ, CIE 1976 LAB, CIECAM, CIELUV, etc. values. This dispenses with the necessity of an expensive spectrometer or spectrophotometer. These values, e.g., of the target are then loaded or entered into a spreadsheet, such as preferably Excel. These entered values are then normalized to a range of 0 to 1 based on the minimum (black) and maximum (white) values.
Next, the value of the light source white point is obtained either through the use of a low cost colorimeter or from the published values of the source. For example, the scanner white point chromaticity values may be obtained from the device vendor. The target white point is then converted to the capture device white point based on simplified CIE Bradford white point adaptation equations. Alternatively, other white point adaptation algorithms known in the art may be used, such as, for example, the full blown Bradford, Von Kries, etc. This will provide the key for the conversion between the target values and those captured from the target. The raw RGB values of the test target are then captured using capture device. It is important that any color management processing, including any gamma correction, of the capture device is disabled during this process. These raw RGB values are then normalized to a range of 0 to 1 based on the minimum (black) and maximum (white) values.
Using the spreadsheet's regression methods (which may be a simple linear regression, or a more complex non-linear, multi-linear, etc. regression) provided by, e.g., Excel's Analysis ToolPak, regress predicted X values based on the normalized X and normalized RGB values without any offset. This results in a 1×3 matrix. This step is repeated for the Y values and the Z values, each time resulting in a 1×3 matrix. These three matrices are then combined into a single 3×3 matrix, from which the computer predicted CIE XYZ values may be obtained. As will be recognized by one skilled in the art, the use of a multi-linear regression can create a single 3×3 matrix in one step instead of the three steps described above. The test target CIE XYZ values in the device white point are used with these predicted CIE XYZ values to compute CIE 1996 color difference values, including average, standard deviation, minimum and maximum delta E*, etc. Based on these values, the predicted accuracy of the capture device is evaluated in the context of its application. Typically an average delta E* of 6 or less is required and 3 or less is desired.
If an acceptable accuracy is achieved, this method then uses the simplified Bradford white point adaptation equations to convert predicted CIE XYZ values from the device white point to D65 (white point of sRGB). As discussed above, the official CIE Bradford equation (including a non-linear aspect for the blue channel) may also be used, or the Von Kries or other algorithms as desired. Next, these values are converted into linear RGB values using the CIE XYZ to RGB matrix from the IEC 61966-2-1 sRGB standard. Finally, these converted values are further converted into sRGB output values using linear to nonlinear tone equations from IEC 61966-2-1 sRGB standard. If an ICC profile is required, the combination of the three 1×3 matrices and a simple tone curve fit of the gray scale data can be used in known fashion. Of course, one skilled in the art will recognize that the output values may be converted to other color spaces as desired. Exemplary additional color spaces, although by no means exhaustive, may be sRGB64, AdobeRGB, etc.
Previous inventions required extensive measurement of the test target to insure the measured white point is identical to the device white point or reworking the device to make the device white point match the published test target white point. Also, most previous inventions ignored nor
Bhat Aditya
Leydig , Voit & Mayer, Ltd.
Microsoft Corporation
LandOfFree
Method of color capture calibration for digital capture devices does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of color capture calibration for digital capture devices, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of color capture calibration for digital capture devices will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3210556