Pattern recognition apparatus and method using probability density function
AssigneeFujitsu Limited
InventorHiroaki Takebe
Life of this patent
Estimating the term…
Abstract
A set of difference vectors is generated by calculating the difference between the feature vector of each pattern in a specific pattern set and the average feature vector of each correct category. When a feature vector of an unknown pattern is inputted, the expected value of the probability density function of a specific category is obtained using an error distribution corresponding to the difference vector set as the probability density function. Then, the discriminant function value for the category is defined based on the obtained expected value and the pattern can be recognized.
Claims10
A pattern recognition apparatus for recognizing a pattern based on a value of a probability density function defined for each category in a feature vector space of a pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition result;
- and
- wherein said calculation device:
- calculating a set of difference vectors by calculating a difference between a feature vector of each pattern in a specific pattern set and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
A pattern recognition apparatus for recognizing a character based on a value of a probability density function defined for each category in a feature vector space of a character pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern of a specific font;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition;
- and
- wherein said calculation device:
- calculating a set of difference vectors by calculating a difference between a feature vector of each character pattern in a character pattern set of the specific font and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
A pattern recognition apparatus for recognizing a character based on a value of a probability density function defined for each category in a feature vector space of a character pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern inputted by specific input equipment;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition;
- and
- wherein said calculation device:
- calculating a set of difference vectors by calculating a difference between a feature vector of each character pattern in a character pattern set, which is inputted by the specific input equipment, and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
A pattern recognition apparatus for recognizing a character based on a value of a probability density function defined for each category in a feature vector space of a character pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern inputted with specific resolution;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition result;
- and
- wherein said calculation device:
- calculating a set of difference vectors by calculating a difference between a feature vector of each character pattern in a character pattern set, which is inputted with the specific resolution, and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
A pattern recognition apparatus for recognizing a character based on a value of a probability density function defined for each category in a feature vector space of a character pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern when a set DSF of difference vectors is generated by calculating a difference between a feature vector of each character pattern in a character pattern set of a specific font F and an average feature vector of each correct category, a set DSI of difference vectors is generated by calculating a difference between a feature vector of each character pattern in a character pattern set, which is inputted by specific input equipment I, and an average feature vector of each correct category, a set DSR of difference vectors is generated by calculating a difference between a feature vector of each character pattern in a character pattern set, which is inputted with specific resolution R, and an average feature vector of each correct category, an expected value f1 of a probability density function f0 of the specific category is defined at each point in the feature vector space using a normal distribution function with a autocorrelation matrix of the set DSF as a covariance matrix, as a probability density function, an expected value f2 of the expected value f1 is defined at each point using a normal distribution with a autocorrelation matrix of the set DSI as a covariance matrix, as a probability density function, an expected value f3 of the expected value f2 is defined at each point using a normal distribution with an autocorrelation matrix of the set DSR as a covariance matrix, as a probability density function, and the discriminant function is defined based on the expected value f3;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition result.
A pattern recognition apparatus for recognizing a character based on a value of a probability density function defined for each category in a feature vector space of a character pattern, comprising:
- a calculation device calculating a value of a discriminant function of a specific category for a feature vector of a target character pattern included in an input document;
- and
- a recognition device recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition result;
- and
- wherein said calculation device:
- calculating a set of difference vectors by calculating a difference between a feature vector of each character pattern in a character pattern set, which are included in the input document and for which a maximum value of a probability density function of a category is larger than a threshold value, and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
A computer-readable storage medium which records a program for enabling a computer to recognize a pattern based on a value of a probability density function that is defined for each category in a feature vector space of a pattern, the program enabling the computer to perform:
- calculating a set of difference vectors by calculating a difference between a feature vector of each pattern in a specific pattern set and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value;
- recognizing the unknown pattern based on the value of the discriminant function;
- and
- outputting a recognition result.
A pattern recognition method for recognizing a pattern based on a value of a probability density function defined for each category in a feature vector space of a pattern, comprising:
- calculating a set of difference vectors by calculating a difference between a feature vector of each pattern in a specific pattern set and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value;
- recognizing the unknown pattern based on the value of the discriminant function;
- and
- outputting a recognition result.
A pattern recognition apparatus for recognizing a pattern based on a value of a probability density function defined for each category in a feature vector space of a pattern, comprising:
- calculation means for calculating a value of a discriminant function of a specific category for a feature vector of an unknown pattern;
- and
- recognition means for recognizing the unknown pattern based on the value of the discriminant function and outputting a recognition result;
- and
- wherein said calculation means:
- calculating a set of difference vectors by calculating a difference between a feature vector of each pattern in a specific pattern set and an average feature vector of each correct category;
- calculating an expected value of a probability density function of a specific category using an error distribution corresponding to the set of difference vectors as the probability density function;
- and
- calculating the value of the discriminant function for the specific category based on the expected value.
Dependencies are inferred from the claim text.