Method of and device for processing image

Claim 2

In a pattern matching image processing method wherein a matching zone in which an image to be read is matched with a templet image is searched and determined, an improvement which comprises the steps of: (a) subjecting a plurality of predetermined picture element areas of said templet image, as blocks, to Hadamard transformation; (b) finding a frequency component of each of a plurality of said blocks giving the maximum absolute Hadamard coefficient value among all frequency components except the D.C. component of each block to find a characteristic templet which has the thus found frequency components as block elements thereof; (c) taking the Hadamard coefficient of said frequency component giving the maximum absolute Hadamard coefficient value as the characteristic value d of each block; (d) finding the characteristic value d' for all blocks of said image to be read while using said characteristic templet; (e) adding the differences of the characteristic values of all images to be read from the characteristic value of the templet to define that the mismatch degree A is represented by A=ε|d-d'|; and (f) determining the matching zone to be the zone at which the mismatch degree assumes a minimum value.

Claim 3

In a pattern matching image processing method wherein a matching zone in which an image to be read is matched with a templet image is searched and determined, an improvement which comprises the steps of: (a) subjecting a plurality of predetermined picture element areas of said templet image as a plurality of blocks to Hadamard transformation; (b) finding the frequency component giving the maximum absolute Hadamard coefficient value among all frequency components of each block except the D.C. components to find a characteristic templet which has the thus found frequency component as a block element thereof; (c) taking the Hadamard coefficient of said frequency component giving the maximum absolute Hadamard coefficient value as the characteristic value d of each of the blocks; (d) finding the characteristic value d' for all blocks of said image to be read while using said characteristic templet; (e) adding the respective differences between the characteristic values of said templet and the characteristic values of said image for the blocks to be read in the order such that the absolute values of said characteristic values d of said templets are larger; (f) comparing the results of the addition of the preceding step at every adding operation with a preset threshold value to define that the read image zone is not a matching zone when the result of addition is larger than said threshold value; and (g) selecting the read image zone giving the minimum added value among all of the read image zone as which give the added values lower than said threshold value and defining the thus selected zone as the matching zone.

Claim 4

In a pattern matching image processing method wherein a matching zone in which an image to be read is matched with a templet image is searched and determined, an improvement which comprises the steps of: (a) subjecting a plurality of predetermined picture element areas of said templet image as a plurality of blocks to Hadamard transformation; (b) finding the frequency component giving the maximum absolute Hadamard coefficient value among all frequency components of each block except the D.C. components to find a characteristic templet which has the thus found frequency component as a block element thereof; (c) taking the Hadamard coefficient of said frequency component giving the maximum absolute Hadamard coefficient value as the characteristic value d of each of the blocks; (d) finding the characteristic value d' for all blocks of said image to be read while using said characteristic templet; (e) adding the respective differences between the characteristic values of said templet and the characteristic values of said image for the blocks to be read in the order such that the absolute values of said characteristic values d of said templets are larger; (f) comparing the results of the addition of the preceding step at every adding operation with a preset threshold value to define that the read image zone is not a matching zone when the result of addition is larger than said threshold value; and (g) selecting the read image zone giving the minimum added value among all of the read image zone as which give the added values lower than said threshold value and defining the thus selected zone as the matching zone.

Claim 5

In an image processing device for processing an image through a pattern matching image processing method wherein a matching zone in which the image to be read is matched with a templet image is searched and determined, an improved device which comprises: a characteristic templet memory for memorizing a characteristic templet which has, as elemental components thereof, the frequency component giving the maximum absolute Hadamard coefficient value among the frequency components except the D.C. component for each area of said templet image; a characteristic value memory for memorizing the Hadamard coefficient of each of the frequency components having the maximum absolute value as the characteristic value d; Hadamard transformer means for transforming said image to be read using said characteristic templet; an additional memory for memorizing the Hadamard coefficients transformed by said transformer means as the characteristic quantity d'; mismatching degree calculating means for calculating a mismatch degree A=ε|d-d'| by adding the absolute values of differences in characterized values of the templet and the image to be read; and comparator means for comparing said mismatch degree with a pre-set value to judge that the zone is a matching zone when said mismatch degree is not more than said pre-set value.

Claim 6

In an image processing device for processing an image through a pattern matching image processing method wherein a matching zone in which the image to be read is matched with a templet image is searched and determined, an improved device which comprises: a characteristic templet memory for memorizing a characteristic templet which has, as elemental components thereof, the frequency component giving the maximum absolute Hadamard coefficient value among the frequency components except the D.C. component for each area of said templet image; a characteristic value memory for memorizing the Hadamard coefficient of each of the frequency components having the maximum absolute value as the characteristic value d; Hadamard transformer means for transforming said image to be read using said characteristic templet; an additional memory for memorizing the Hadamard coefficients transformed by said transformer means as the characteristic quantity d'; mismatching degree calculating means for calculating a mismatch degree A=ε|d-d'| by adding the absolute values of differences in characterized values of the templet and the image to be read; and comparator means for comparing said mismatch degree with a pre-set value to judge that the zone is a matching zone when said mismatch degree is not more than said pre-set value.

Claim 7

In an image processing device for processing an image through a pattern matching image processing method wherein a matching zone in which the image to be read is matched with a templet image is searched and determined, an improved device which comprises: a characteristic templet memory for memorizing a characteristic templet which has, as elemental components thereof, the frequency component giving the maximum absolute Hadamard coefficient value among the frequency components except the D.C. component for each area of said templet image; a characteristic value memory for memorizing the Hadamard coefficient of each of the frequency components having the maximum absolute value as the characteristic value d and simultaneously memorizing at least the area giving the maximum absolute value; Hadamard transformer means for transforming said image to be read using said characteristic templet; calculator means for calculating and adding the absolute values of the differences in characteristic values of the templet and of the image to be read in the order from the area which assumes the maximum characteristics value; and comparator means for comparing the sum obtained by said calculator with a pre-set threshold to judge that the image is mismatched when said sum is more than said pre-set threshold.

Claim 8

In an image processing device for processing an image through a pattern matching image processing method wherein a matching zone in which the image to be read is matched with a templet image is searched and determined, an improved device which comprises: a characteristic templet memory for memorizing a characteristic templet which has, as elemental components thereof, the frequency component giving the maximum absolute Hadamard coefficient value among the frequency components except the D.C. component for each area of said templet image; a characteristic value memory for memorizing the Hadamard coefficient of each of the frequency components having the maximum absolute value as the characteristic value d and simultaneously memorizing at least the area giving the maximum absolute value; Hadamard transformer means for transforming said image to be read using said characteristic templet; calculator means for calculating and adding the absolute values of the differences in characteristic values of the templet and of the image to be read in the order from the area which assumes the maximum characteristics value; and comparator means for comparing the sum obtained by said calculator with a pre-set threshold to judge that the image is mismatched when said sum is more than said pre-set threshold.