Color measuring system using multispectral image data

Details Color measuring system using multispectral image data Color measuring system using multispectral image data

1996-05-16

2001-07-31

Chang, Jon (Department: 2623)

Image analysis

Color image processing

Pattern recognition or classification using color

C382S227000, C356S418000

Reexamination Certificate

active

06269182

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a color measuring system that measures the colors of objects using multispectral image data obtained via a plurality of band-pass filters, and more particularly to a color classifying apparatus that sorts objects by color and a color nonuniformity checking apparatus that checks objects for color nonuniformity.
2. Description of the Related Art
Color identifying systems that identify the color of objects and color measuring systems that measure the colors of objects have been used in managing both painting and dyeing at job sites in various industries as well as for measuring the colors of specimens in the medical and scientific fields.
Color identifying systems and color measuring systems of this type have been disclosed in, for example, Jpn. Pat. Appln. KOKAI Publication No. 3-267726, where the reflected spectral spectrum of an object is sorted into two classes by performing a statistical process on the spectrum.
Specifically, the reflected spectral spectrum of an object whose class is known is subjected to a statistical process using the Foley Sammon transform (FS transform) method (refer to Q. Tian, M. Barbaro, et al., “Image classification by the Foley-Sammon transform,” Optical Engineering, Vol. 25, No. 7, 1986).
The FS transform method is an approach whereby a spectrum is sorted into two classes. Specifically, this approach is to find spectrum di for a classification that maximizes Fisher ratio R(di) determined by the following equation (1):
R
(
di
)=(
di
t
S
1
di
)/(
di
t
S
2
di
)   (1)
where di is a classification spectrum,
di
t
is a classification spectrum (transposition),
S
1
is an interclass covariance matrix and
S
2
is an intraclass covariance matrix.
Hereinafter, spectrum di for this classification is referred to as a classification spectrum.
Since the classification spectrum di has the same number of dimensions as the spectrum of the object, to be precise, it should be expressed as di(&lgr;), but for the sake of simplicity, it is represented as di.
Two types of classification spectrums that make the Fisher ratio larger will be found.
A classification spectrum di that maximizes the Fisher ratio is determined to be d
1
and the one that maximizes the Fisher ratio in the spectrums perpendicular to d
1
is determined to be d
2
.
By projecting each data item on the space constructed by the two classification spectrums d
1
and d
2
, a spectrum is sorted into two classes.
The classification spectrums d
1
and d
2
are obtained from the following equation (2):
d
1
=&agr;
1
S
2
−1
&Dgr;,
d
2
=&agr;
2
S
2
−1
[I
−(&Dgr;
t
S
2
−2
&Dgr;)/(&Dgr;
t
S
2
−3
&Dgr;)
S/
2
−1
]&Dgr;  (2)
where &agr;
1
and &agr;
2
are normalization coefficients, &Dgr; is X
1
−X
2
(the differential spectrum between class 1 and class 2), and I is a unit matrix.
To project each data item on the space composed of the classification spectrums d
1
and d
2
thus obtained, the inner product of the classification spectrum and the reflected spectral spectrum of the object is found.
If the reflected spectral spectrum of the object is expressed as f(&lgr;) (&lgr;=wavelength), the inner products t
1
and t
2
are expressed by the following equation (3):
t
1
=
f
(&lgr;)·
d
1
t
2
=
f
(&lgr;)·
d
2
  (3)
where · indicates inner product operation.
In the prior art as disclosed in Jpn. Pat. Appln. KOKAI Publication No. 3-267726, a classification boundary is determined from the values of t
1
and t
2
as shown in
FIG. 32 and a
filter having the spectrum characteristic is realized using a diffraction grating and a liquid-crystal filter
2
as shown in FIG.
33
.
In
FIG. 33
, reference numeral
3
indicates a light-source lamp.
The classification spectrums d
1
and d
2
are generally complex in shape as shown in FIG.
34
and the accuracy with which the diffraction grating
1
and liquid-crystal filter
2
are mounted must be high because the spectrums take positive and negative values.
The prior-art system as disclosed in Jpn. Pat. Appln. KOKAI Publication No. 3-267726 has disadvantages in that: (i) because the system restricts the light source to a particular type, it is not suited for classification with a different light source and cannot achieve effective classification when the spectrum of the light source varies; and (ii) the diffraction grating used is expensive.
To overcome these disadvantages, the assignee of the present invention has filed Japanese Patent Application No. 6-241614, which discloses an invention related to a color classifying apparatus that classifies the colors of objects using the multispectral images obtained via a plurality of band-pass filters. This system has a simplified configuration, reduced cost, and better resistance to mechanical vibration, and is also capable of classifying the colors of objects well without limiting the light source even when its spectrum varies.
As a color nonuniformity checking apparatus that checks objects for color nonuniformity, a spectroscope or a color difference meter have been used. In addition, color nonuniformity has been checked on the basis of the RGB input from a color video camera.
In the color classifying apparatus disclosed in Japanese Patent Application No. 6-241614, however, since classification judgment is made by only one type of classification judgment method, the apparatus has left much room for improvement with respect to the accuracy of classification.
Specifically, with only one type of classification judgment method, the performance of classification judgment decreases seriously, depending on the state of the distribution in a multidimensional space of a plurality of objects subjected to the classification judgment. The classification performance tends to deteriorate, depending on the object.
Furthermore, when an object is checked for color nonuniformity with a conventional spectroscope or a color difference meter, it is impossible to check the object for color nonuniformity through one measurement because only spot measurement can be made.
In this case, if measurement is made several times, variations will take place in each measurement, so that the color nonuniformity of the object cannot be determined accurately.
Furthermore, when color nonuniformity is checked for on the basis of the RGB input from a color video camera, it is difficult to sense a subtle color difference because of the characteristics of the color filter used, so that the color nonuniformity of the object cannot be checked accurately.
SUMMARY OF THE INVENTION
Accordingly, an object of the present invention is to provide a color classifying apparatus which performs the process of classifying the colors of objects using the multispectral images obtained via a plurality of band-pass filters, simplifies the configuration, reduces cost, withstands mechanical vibration, and is capable of classifying the colors of objects well without limiting the light source even when its spectrum varies, and which is capable of further improving the classification accuracy by using a classification judgment method most suitable for making a classification judgment on objects.
Another object of the present invention is to provide a color nonuniformity checking apparatus which performs the process of checking objects for color nonuniformity using the multispectral images obtained via a plurality of band-pass filters, simplifies the configuration, reduces cost, withstands mechanical vibration, and is capable of checking objects for color nonuniformity well without limiting the light source even when its spectrum varies, and which is capable of improving the accuracy in checking object for color nonuniformity.
The foregoing object is accomplished by providing a color classifying apparatus comprising: imaging means for imaging the reflected light from an object; optical means for forcing the reflected light from the object to form multispectral images each having a different band on t

Affiliated with

Also associated with

Rating

Color measuring system using multispectral image data does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Color measuring system using multispectral image data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Color measuring system using multispectral image data will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2550786