Data processing: measuring – calibrating – or testing – Calibration or correction system – Sensor or transducer
Reexamination Certificate
2001-05-04
2003-03-25
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Calibration or correction system
Sensor or transducer
C702S085000, C356S402000, C356S408000
Reexamination Certificate
active
06539323
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates generally to the measurement of color, and more particularly to the correction of spectral measurements, such as are obtained by a spectrophotometer for measuring color.
Accurate and reliable measurement of color is difficult to achieve, but is essential for successful color reproduction and control. “Color” is the human visual system's response to a specific distribution of light energy in what is termed the “visible spectrum”, a portion of the full electromagnetic energy spectrum where the wavelength ranges from 400 nanometers (1 nm =10
−9
meters) to 700 nm. A plot of light energy versus wavelength is called a spectrum. As an example, a spectrum for the color Magenta is shown in FIG.
1
. The various possible shapes of the spectrum plot gives rise to the perception of different colors. For example, spectra which have larger energy amplitudes in the short wavelengths near 400 nm are perceived as being “blue” while spectral plots that show larger amplitudes in the longer wavelengths near 700 nm are perceived as being “red”. The set of all possible shapes for a spectral energy plot gives rise to the enormous number of colors that humans can see (nearly 10 million). In the case of visible light, each spectral band of color may be as small as 2 nm wide spanning the range from 400 nm to 700 nm.
A wide variety of instruments are used to make quantitative measurements of color. In general, these instruments can be classified as making either “three-band”, or “full spectrum” measurements. The three-band instruments measure the light energy reflected from a sample at three positions within the spectrum. They are not able to detect the entire spectrum of a color, but having a 3-channel estimate of it is very useful for many printing and color control applications, and the expense of making this kind of measurement is low.
Full spectrum instruments are able to obtain the spectral energy distribution of a color across the entire visible spectrum, and thereby gain a more accurate representation of the color characteristics of a sample. For example, such a measurement can be used to predict a sample's color appearance even when the lighting on the sample changes.
The more accurately a spectrum is measured, the better will be the color representation, and so very high resolution spectra (many sampling positions along the wavelength axis) are desirable. It is difficult however, to make accurate and high resolution full spectrum measurements.
For a variety of reasons related to the specific design of an instrument, the amplitude at one position (one wavelength) along the spectrum is influenced by the amplitudes at other wavelengths. This is called “cross-spectrum contamination” or “crosstalk”. If the amplitude measured at some position in the spectrum is distorted by crosstalk, the spectrum will misrepresent the color of the sample.
FIG. 2
illustrates an example of this effect.
FIG. 2
shows a spectrum of a blue sample. The plot of actual spectra
10
shows that the reflected energy from the sample is highest in the short wavelength region and the spectrum makes a transition to a low level for the rest of the wavelengths. A plot of a spectrum that might be measured by an instrument suffering from spectral crosstalk is shown at
12
. Plot
12
shows a rise in energy at the long wavelength end that does not really exist, it is a false measurement of the blue energy showing up as an apparent amount of red. In other words, plot
12
shows energy from the blue end of the spectrum being falsely detected as energy from the red end of the spectrum. The color represented by the measurement will have a reddish tint compared to the actual color.
If it could be determined just how much the blue wavelength energy was influencing the red wavelength measurements, we could compensate for this effect. Unfortunately, a single measurement isn't enough to discover exactly what wavelength is causing the contamination. The contamination could be any of the wavelengths in the blue region, it could be a little contamination from all of the wavelengths, the contamination could be from just a portion of the wavelengths, or from a gradual increase of contamination toward one specific wavelength.
To find out the details of the crosstalk in order to correct for it, many measurements must be taken of unique color spectra, and then the influencing wavelengths must be factored out. The number of measurements that must be made is equal to the number of wavelength positions along the spectrum that are used in the spectral plot. It is common to have at least 30 such positions, but more accurate instruments using higher resolution obtain over 100 individual amplitudes for a spectral plot. It becomes difficult to solve for the crosstalk characteristics for such a large system.
In addition, the nature of the light source used to illuminate a color specimen can have considerable effect on the spectrum that is detected. In particular, the angle of illumination, the texture and gloss of the sample, and the amount of ultraviolet energy in the light and fluorescent material in the sample, all influence the spectrum that is obtained.
Because of these issues, prior art spectral sensing instruments incorporate many different strategies and detection techniques. Any spectral instrument design represents a collection of tradeoffs involving sensitivity, accuracy, specimen size and geometry, cost, power consumption, and the like. These tradeoffs result in considerable variation in instrument designs available on the market, and correspondingly, variations in the color measurements made by them.
It would be advantageous to provide a simple means for correcting spectral color measurements using vectors and matrices. It would be advantageous to provide for the correction of crosstalk effects as well as various other sources of spectral representation error. It would be further advantageous to provide for the correction of spectral color measurements using a single correction matrix. It would be still further advantageous to provide a correction matrix that embodies a transform that minimizes the difference between the corrected spectra and a set of reference spectra. It would be advantageous if the difference between the corrected spectra and the reference spectra could be characterized by a set of basis functions, which can be used to build the correction matrix.
The methods and apparatus of the present invention provide the aforesaid and other advantages.
GENERAL DESCRIPTION OF THE INVENTION
The present invention provides methods and apparatus for correcting spectral measurements, such as are obtained by a spectrophotometer for measuring color. In accordance with the invention, a single matrix operates on a raw measurement vector (spectrum) to obtain a corrected spectrum. The matrix may embody a transform that minimizes the difference between the corrected spectra and a set of reference spectra. The difference may be characterized by a set of basis function weighting vectors which are then used to build the correction matrix. The method allows the correction of high resolution spectra (very long measurement vectors) without requiring the large number of measurements that would normally be required. The reference spectra can be calibration data, or measurements made by another instrument which is desired to be simulated.
This invention provides a way to solve for the crosstalk components without making hundreds of measurements for a high resolution full spectrum color instrument. Instead, a few tens of measurements can be made. The (uncalibrated) color instrument to be characterized is used to obtain spectral measurements of, for example, 24 uniquely colored sample patches. Another instrument, a reference instrument, which is known to be calibrated accurately, also makes measurements and obtains spectra of these same 24 color patches. The spectra from the reference instrument are collected and compared to the set of spectra from the uncalibrated instrumen
Barlow John
Electronics For Imaging, Inc.
Lipsitz Barry R.
McAllister Douglas M.
Vo Hien
LandOfFree
Methods and apparatus for correcting spectral color... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for correcting spectral color..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for correcting spectral color... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3011546