Image analysis – Pattern recognition – Classification
Reexamination Certificate
1998-08-24
2002-05-28
Boudreau, Leo (Department: 2621)
Image analysis
Pattern recognition
Classification
C382S116000, C382S118000, C382S159000, C382S160000, C382S190000, C382S209000, C382S284000, C707S793000, C707S793000, C707S793000
Reexamination Certificate
active
06396954
ABSTRACT:
TECHNICAL FIELD
The present invention relates to an apparatus and a method for recognition and an apparatus and a method for learning. More particularly, the present invention relates to an apparatus and a method for recognition and an apparatus and a method for learning in which in recognizing e.g. sounds and objects, other data is utilized as well as their audio and video data to increase the recognition accuracy.
BACKGROUND ART
In a conventional voice recognition apparatus for recognizing voice sounds, voice data picked up by a microphone is (acoustically) analyzed and an analyzed result is used to recognize the voice emitted by a user.
However, such a conventional voice recognition apparatus utilizes the analyzed result from the voice data picked up by the microphone for voice recognition, whereby its recognition accuracy will be limited to a certain level.
It should be understood that not only the voice data picked up by a microphone but also other factors such as the expression and the movement of the mouth of a subject are notable and thus concerned for recognizing the voice of the subject.
The voice recognition apparatus is normally used under hostile conditions where different types of noise are received but not in a particular circumstance, e.g. a sound-proof chamber, where the voice of a subject only can be picked up by a microphone. In particular, a renovated navigation system may be equipped with such a voice recognition apparatus which however receives unwanted noise sounds, including sounds of a CD (compact disk) player, an engine, and an air-conditioner mounted in a vehicle, other than the voice of a subject to be recognized. Since it is very difficult to remove noise sounds from the voice data, the voice recognition has to deal with the noise sounds for improving its accuracy.
It is also common in the conventional voice recognition apparatus that the voice data picked up by a microphone is processed by a specific manner to determine characteristic parameters and the voice recognition is carried out by calculating the distance between the characteristic parameters plotted in a parameter space. As a rule, the characteristic parameters which are essential for the voice recognition are varied depending on the conditions where the voice recognition apparatus is set.
DISCLOSURE OF THE INVENTION
The present invention is directed towards overcoming the foregoing drawbacks and its object is to increase the recognition accuracy of a recognition apparatus for recognizing voice or other factors.
A recognition apparatus, as defined in claim
1
, comprises: a first classifying means for classifying different types of input data into classes depending on their characteristics; an integrated parameter constructing means for constructing an integrated parameter through integrating the different types of input data; a standard parameter saving means for saving tables, each table carrying standard parameters and assigned to one of the classes determined by the first classifying means; and a recognizing means for recognizing a given subject using the integrated parameter and the standard parameters listed in the table assigned to the class determined by the first classifying means.
A recognition method, as defined in claim
5
, comprises the steps of: classifying different types of input data into classes depending on their characteristics and constructing an integrated parameter through integrating the different types of input data; and recognizing a given subject using the integrated parameter and a table carrying standard parameters and assigned to one of the classes determined by the classification.
A learning apparatus, as defined in claim
6
, comprises: a first classifying means for classifying different types of input data into classes depending on their characteristics; an integrated parameter constructing means for constructing an integrated parameter through integrating the different types of input data; and a classifying means for classifying the integrated parameters according to the class determined by the first classifying means.
A learning method, as defined in claim
9
, comprises the steps of: classifying different types of input data into classes depending on their characteristics and constructing an integrated parameter through integrating the different types of input data; and classifying the integrated parameters according to the class determined by the classification.
In the recognition apparatus defined in claim
1
, the first classifying means classifies the different types of input data into classes depending on their characteristics and also, the integrated parameter constructing means constructs an integrated parameter through integrating the different types of input data. The standard parameter saving means includes tables, each table carrying standard parameters and assigned to one of the classes determined by the first classifying means. The recognizing means thus recognizes a given subject using the integrated parameter and the standard parameters listed in the table assigned to the class determined by the first classifying means.
In the recognition method defined in claim
5
, different types of input data are classified into classes depending on their characteristics and an integrated parameter is constructed through integrating the different types of input data. Then, a given subject can be recognized using the integrated parameter and a table carrying standard parameters and assigned to one of the classes determined by the classification.
In the learning apparatus defined in claim
6
, the first classifying means classifies different types of input data into classes depending on their characteristics and the integrated parameter constructing means constructs an integrated parameter through integrating the different types of input data. The classifying means also classifies the integrated parameters according to the class determined by the first classifying means.
In the learning method defined in claim
9
, different types of input data are classified into classes depending on their characteristics and an integrated parameter is constructed by integrating the different types of input data. The integrated parameters are then classified according to the class determined by the classification.
REFERENCES:
patent: 4825362 (1989-04-01), Minami et al.
patent: 4937869 (1990-06-01), Iwahashi et al.
patent: 4965725 (1990-10-01), Rutenberg
patent: 4975969 (1990-12-01), Tal
patent: 4991216 (1991-02-01), Fujii et al.
patent: 5522011 (1996-05-01), Epstein
patent: 5561718 (1996-10-01), Trew et al.
patent: 5749072 (1998-05-01), Mazurkiewicz et al.
patent: 5754681 (1998-05-01), Watanabe et al.
patent: 5761329 (1998-06-01), Chen et al.
patent: 5859925 (1999-01-01), Yaeger et al.
patent: 5956463 (1999-09-01), Patrick et al.
patent: 5983186 (1999-11-01), Miyazawa et al.
patent: 6023529 (2000-02-01), Ilan et al.
patent: 6064323 (2000-05-01), Ishii et al.
patent: 44 36 692 (1995-04-01), None
patent: 2-262199 (1990-10-01), None
patent: 3-129400 (1991-06-01), None
patent: 06-301393 (1994-10-01), None
patent: 07-028490 (1995-01-01), None
patent: 08-187368 (1996-07-01), None
patent: 9-134193 (1997-05-01), None
Wagner T. et al: “Multi-Sensorial Inputs For The Identification Of Persons With Synergetic Computers” Proceedings of the International Conference On Image Processing (ICIP), US Los Alamitos, IEEE Comp. Soc. Press, vol. Conf. 1, 1994, pp. 287-291 XP000522640.
Silbee P. L.: “Sensory Integration In Audiovisual Automatic Speech Recognition” Record of the Asilomar Conference On Signals, Systems and Computers, US, Los Alamitos, IEEE Comp> Soc. Press, vol. Conf. 28, 1995, pp. 561-565 XP000533615.
Jennings D. L. et al: “Enhancing Automatic Speech Recognition With An Utrasonic LIP Motion Detector” Proceedings of the International Conference On Acoustics. Speech And Signal Processing (ICASSP), US, New York, IEEE, 1995, pp. 868-871 XP000658132.
Boudreau Leo
Frommer William S.
Frommer & Lawrence & Haug LLP
Mariam Daniel G.
Simon Darren M.
LandOfFree
Apparatus and method for recognition and apparatus and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for recognition and apparatus and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for recognition and apparatus and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2855787