Data classifying apparatus and material recognizing apparatus

1. A data classifying apparatus comprising:

a class constructor that classifies model data groups, each comprising a plurality of model data and represented in a vector format, into classes on a per category basis, said classes comprising smaller vector distributions than categories required by an application, each class representing characteristics in more detail than said categories;

a representative value calculator that calculates a representative value of each class using the model data classified in each respective class;

a class determiner that compares input data to the representative values, and determines a class having the representative value most similar to the input data; and

a category determiner that determines a category to which the input data belongs from the determined class.

2. The apparatus according to claim 1 , wherein the distribution of each class does not belong to a plurality of categories.

3. The apparatus according to claim 1 , further comprising:

a matrix calculator that calculates a characteristic extraction matrix that transforms a coordinate system so as to clarify characteristics of each class, wherein the category determiner compares the input data with representative values transformed by the characteristic extraction matrix and thereby determines the class having the representative value most similar to the input data.

4. The apparatus according to claim 1 , wherein the class constructor classifies the model data groups into classes so as to minimize variances within the classes.

5. The apparatus according to claim 1 , wherein the data classifying apparatus generates, from the model data, a model pattern vector expressed as a one dimensional vector representative of pixel values and generates, from the input data an input pattern vector expressed as a one dimensional vector representative of pixel values.

6. The apparatus according to claim 2 , wherein when the category determiner determines that the input data belongs to a plurality of categories, the class constructor adds a new item to the predetermined item, and classifies the classes according to the new item, thereby preventing a distribution of any of the classes from belonging to a plurality of categories.

7. The apparatus according to claim 3 , wherein the matrix calculator calculates the characteristic extraction matrix so as to decrease a variance of each of the model data groups within each of the classes.

8. The apparatus according to claim 3 , wherein the matrix calculator transforms the characteristic extraction so as to increase a variance of the classes between the classes.

9. The apparatus according to claim 3 , wherein the matrix calculator calculates the characteristic extraction matrix so as to decrease a variance of each of the model data groups within each of the classes, while increasing a variance of the classes between the classes.

10. The apparatus according to claim 9 , wherein the matrix calculator calculates the characteristic extraction matrix so as to maximize a variance ratio of a between-class covariance matrix of the classes to a within-class covariance matrix of the model data groups contained in the classes.

11. The apparatus according to claim 9 , wherein the matrix calculator calculates the characteristic extraction matrix so as to maximize a variance ratio of a covariance matrix of the model data groups to a within-class covariance matrix of the model data groups contained in the classes.

12. The apparatus according to claim 11 , wherein the class constructor transforms the coordinate system such that covariance matrices of the model data groups are unit matrices.

13. A material recognizing apparatus, comprising:

the data classifying apparatus according to claim 1 ;

wherein the category comprises a type of material, the model data is classified according to a type, and the type of the input data is determined.

14. A material recognizing apparatus, comprising:

the data classifying apparatus according to claim 1 ;

wherein the category comprises a distance to a material, the model data is classified according to a distance to the material, and the distance to the material of the input data is determined.

15. A data classifying method, comprising:

classifying model data groups, each comprising a plurality of model data and represented in a vector format, into classes on a per category basis, the classes comprising smaller vector distribution than categories required by an application, each class representing characteristics in more detail than the categories;

calculating a representative value of each class using the model data classified in each respective class;

comparing input data to the representative values;

determining a class having the representative value most similar to the input data; and

determining a category to which the input data belongs from the determined class.

16. A program executable by a computer, comprising:

classifying model data groups, each comprising a plurality of model data and represented in a vector format, into classes on a per category basis, the classes comprising smaller vector distribution than categories required by an application, each class representing characteristics in more detail than the categories;

calculating a representative value of each class using the model data classified in each respective class;

comparing input data to the representative values;

determining a class having the representative value most similar to the input data; and

determining a category to which the input data belongs from the determined class.