Pattern feature extraction via fourier amplitudes of a block image
An input pattern feature amount is decomposed into element vectors. For each of the feature vectors, a discriminant matrix obtained by discriminant analysis is prepared in advance. Each of the feature vectors is projected into a discriminant space defined by the discriminant matrix and the dimensions are compressed. According to the feature vector obtained, projection is performed again by the discriminant matrix to calculate the feature vector, thereby suppressing reduction of the feature amount effective for the discrimination and performing effective feature extraction.
1. A pattern feature extraction method comprising:
dividing via a processor an input image into sets of a plurality of equally partitioned block images, wherein each dividing step divides said input image into said sets of said plurality of equally partitioned block images by a divisor, the divisor for each set differing from one set to another;
extracting via a processor a Fourier amplitude of each of the input image and the block images that are obtained in each dividing step for generating a multiblock Fourier amplitude vector composed of the thus extracted Fourier amplitudes of the input image and the block images to thereby extracting a feature amount of the input image;
projecting the multiblock Fourier amplitude vector using a basis matrix to obtain a projection vector; and
normalizing the projection vector to obtain a normalized vector,
wherein the basis matrix is a basis matrix specified by a transformation matrix for extracting principal components of the multiblock Fourier amplitude vector and by a discriminant matrix obtained by discriminant analysis on the principal components.