Method and apparatus for extracting line segments from an image of an object

Claim 2

A computer-implemented method of extracting a line segment contained in an original image, comprising the steps of: inputting an original two-dimensional image of an object; clipping an image, which corresponds to the size of a receptive field defining a small area, from the original two-dimensional image; projecting the clipped receptive-field image from a predetermined projection direction onto an axis (ρ axis) that intersects said projection direction and obtaining projection values at positions along the ρ axis; subjecting the projecting values to differentiation processing along the ρ axis; similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane; extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ ) of the differential values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 3

The computer-implemented method according to claim 1, wherein the original image is subdivided into receptive fields, said line-segment extraction processing is applied to each receptive-field image, and a line segment contained in the original image is extracted.

Claim 4

The computer-implemented method according to claim 1, wherein a step of calculating a projection value at the projection-direction angle θ includes steps of: setting a projection grid having sides parallel to a and ρ axes on a receptive field, where an axis along the projection axes is the a axis and an axis perpendicularly intersecting the a axis is the ρ axis; accumulating image data of pixels, of the receptive field, that are overlapped by each pixel within the projection grid for which the value along the ρ axis is ρ 0 ; and calculating, as a projection value that prevails when the projection-direction angle is θ, each accumulated value obtained by varying ρ 0 incrementally a predetermined amount up to the width of the projection grid.

Claim 5

The computer-implemented method according to claim 3, wherein a center of the projection grid and a center of the receptive field are made to coincide, the projection grid is successively rotated incrementally a predetermined amount and projection values in all projection directions are calculated.

Claim 6

The computer-implemented method according to claim 4, wherein all projection directions range from 0° to 180°.

Claim 7

The computer-implemented method according to claim 3, wherein there is provided coordinate transformation means for coordinate-transforming pixel positions within the projection grid into pixel positions within the receptive field, pixel positions, within the projection grid, for which the value along the ρ axis is ρ 0 are subjected to a coordinate transformation to obtain corresponding pixel positions within the receptive field.

Claim 8

The computer-implemented method according to claim 3, wherein the size of the receptive-field is larger than the size of the projection grid.

Claim 9

The computer-implemented method according to claim 3, further including the steps of: modifying shape of the projection grid in such a manner that the number of pixels accumulated gradually diminishes as the edge of said grid is approached from the center thereof; and dividing the accumulated value of the image data at each position along the ρ axis by the number of pixels accumulated, thereby calculating projection values by performing normalization.

Claim 10

The computer-implemented method according to claim 3, further including steps of: reducing weighting of pixels as the periphery of said receptive field is approached from the center thereof; obtaining a weighted image by multiplying each item of pixel data of the receptive-field image by a weighting; and calculating projection values using said weighted value.

Claim 11

The computer-implemented method according to claim 3, further including steps of: enlarging degree of vignetting of pixels as the periphery of said receptive field is approached from the center thereof; obtaining a vignetted image by applying vignetting processing to each item of pixel data of the receptive-field image; and calculating projection values using said vignetted image.

Claim 12

The computer-implemented method according to claim 1, further including steps of: detecting a peak position (ρ,θ) having a maximum value among said differential values; and extracting a line segment contained in the receptive-field image based upon said peak position (ρ,θ).

Claim 13

The computer-implemented method according to claim 1, further including steps of: comparing a differential value at a predetermined position on the ρ-θ plane and differential values at surrounding positions; and adopting the predetermined position as the peak position if the differential value at said predetermined position is maximum.

Claim 14

The computer-implemented method according to claim 11, further including steps of: deciding a function, in which a variable is ρ or θ, that gives the differential value at the peak position and a differential value at a position adjacent to said peak position along the ρ or θ axis; and adopting the peak position of said functional value as a true peak position.

Claim 15

The computer-implemented method according to claim 2, further including steps of: subdividing the original image into partial areas in which numbers of pixels in horizontal and vertical directions are p 1 ·m, p 2 ·m (where p 1 , p 2 represent integers of 2 or more and m, n represent numbers of pixels in the horizontal and vertical directions of the receptive field); taking m pixels per p 1 pixels from the partial areas in the horizontal direction and n pixels per p 2 pixels from the partial areas in the vertical direction; and performing line-segment extraction using the m pixel and n pixels as the receptive-field image; and performing line-segment extraction upon successively reading the values of p 1 , p 2 .

Claim 16

A computer-implemented method of extracting a line segment contained in an original image, comprising the steps of: inputting an original two-dimensional image of an object; adopting a small area comprising m×n pixels as a receptive field, projecting one receptive-field image, which has been clipped from the original two-dimensional image, from a prescribed projection direction upon the axis (ρ axis) perpendicular to the projection direction to obtain projection values, and subjecting said projection values to differentiation processing along the ρ axis to obtain a differential value at a center position ρc along the ρ axis; similarly applying differentiation processing, along the ρ axis, to projection values obtained while varying the angle θ of the projection direction successively by a prescribed incremental amount, thereby obtaining differential values at the center position ρc along the ρ axis in all projection directions; shifting a position (h,v) of the receptive-field image, which is clipped from the original image, one pixel at a time in the horizontal and/or vertical direction; obtaining a differential value at the center position ρc along the ρ axis in all projection directions of each receptive field; extracting a line segment contained in the receptive-field images based upon the peak positions (ρc, θ) of the differential values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 17

A computer-implemented method of extracting a line segment contained in an original image, comprising the steps of: inputting an original two-dimensional image of an object; adopting a small area comprising m×n pixels as a receptive field, projecting one receptive-field image, which has been clipped from the original two-dimensional image, from a prescribed projection direction upon the axis (ρ axis) perpendicular to the projection direction to obtain projection values, and subjecting said projection values to differentiation processing along the ρ axis to obtain a differential value at a center position ρc along the ρ axis; similarly applying differentiation processing, along the ρ axis, to projection values obtained while varying the angle θ of the projection direction successively by a prescribed incremental amount, thereby obtaining differential values at the center position ρc along the ρ axis in all projection directions; shifting a position (h,v) of the receptive-field image, which is clipped from the original image, one pixel at a time in the horizontal and/or vertical direction; obtaining a differential value at the center position ρc along the ρ axis in all projection directions of each receptive field; extracting a line segment contained in the receptive-field images based upon the peak positions (ρc, θ) of the differential values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 18

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto an axis (ρ axis) that perpendicularly intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; setting a projection grid, which has sides parallel to an a axis and the ρ axis, on the receptive field of the original two-dimensional image, in which the a axis is an axis in the projection direction; accumulating image data of pixels, of the receptive field, corresponding to respective ones of a number of pixels within said projection grid for which the value along the ρ axis is ρ0; calculating, as projection values that prevail when the projection-direction angle is θ, accumulated values, at positions along the ρ axis, obtained by varying ρ0 incrementally a predetermined amount up to the width of the projection grid; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment; said step of accumulating the image data includes steps of: providing coordinate transformation means for coordinate-transforming a pixel position within the projection grid into a pixel position within the receptive field, and obtaining a position within the receptive field by applying a coordinate transformation to a pixel position, within the projection grid, for which the value along the ρ axis is ρ0; obtaining a weighted mean value of each item of pixel data indicative of a pixel within the receptive field belonging to said position and a pixel adjacent to said pixel; calculating said weighted mean value with respect to each pixel position, within the projection grid, for which the value along the ρ axis is ρ0; and calculating a projection value, for which the value along the ρ axis is ρ0, by accumulating the weighted mean value obtained.

Claim 19

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto an axis (ρ axis) that perpendicularly intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; setting a projection grid, which has sides parallel to an a axis and the ρ axis, on the receptive field of the original two-dimensional image, in which the a axis is an axis in the projection direction; accumulating image data of pixels, of the receptive field, corresponding to respective ones of a number of pixels within said projection grid for which the value along the ρ axis is ρ0; calculating, as projection values that prevail when the projection-direction angle is θ, accumulated values, at positions along the ρ axis, obtained by varying ρ0 incrementally a predetermined amount up to the width of the projection grid; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment; said step of accumulating the image data includes steps of: providing coordinate transformation means for coordinate-transforming a pixel position within the projection grid into a pixel position within the receptive field, and obtaining a position within the receptive field by applying a coordinate transformation to a pixel position, within the projection grid, for which the value along the ρ axis is ρ0; obtaining a weighted mean value of each item of pixel data indicative of a pixel within the receptive field belonging to said position and a pixel adjacent to said pixel; calculating said weighted mean value with respect to each pixel position, within the projection grid, for which the value along the ρ axis is ρ0; and calculating a projection value, for which the value along the ρ axis is ρ0, by accumulating the weighted mean value obtained.

Claim 20

The computer-implemented method according to claim 16, wherein said coordinate transformation means has a ROM for storing, for each and every projection direction, data indicating a position, within the receptive field, conforming to each pixel position (ρ,θ) within the projection grid.

Claim 21

The computer-implemented method according to claim 16, wherein said weighted mean value is calculated upon weighting the image data of each pixel based upon a distance from a position, within the receptive field, obtained by said coordinate transformation to centers of adjacent pixels within the receptive field.

Claim 22

The computer-implemented method according to claim 16, further including steps of: setting first and second parallel lines at positions on both sides of a projection center line at a predetermined distance from the projection center line, wherein said projection center line connects the centers of pixels, within the projection grid, for which the value along the ρ axis is ρ 0 ; and calculating said weighted mean value by weighting the image data of each pixel in conformity with distances from a boundary line between said adjacent pixels within the receptive field to said first and second parallel lines.

Claim 23

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto an axis (ρ axis) that perpendicularly intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; setting a projection grid, which has sides parallel to an a axis and the ρ axis, on the receptive field of the original two-dimensional image, in which the a axis is an axis in the projection direction; coordinate-transforming each pixel position within the projection grid, for which the value along the ρ axis is ρ0, into pixel positions within the receptive field; accumulating pixel data of each pixel, within the receptive field, obtained by the coordinate transformation; calculating, as projection values that prevail when the projection-direction angle is θ, accumulated values, at positions along the ρ axis, obtained by varying ρ0 incrementally a predetermined amount up to the width of the projection grid; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment; said coordinate transformation step including steps of: storing, in correlation with position along the ρ axis, a y address (where a pixel position within the receptive field is expressed by an x-y address) of a pixel group, within the receptive field, that corresponds to a pixel group of a=0 among pixels within said projection grid, and storing said correlation, in a first memory, for each and every projection direction; storing, in correlation with the y address, an x address of a pixel group, within the receptive field, that corresponds to a pixel group of ρ=0 among pixels within said projection grid, and storing said correlation, in a second memory, for each and every projection direction; and calculating an x, y address of each pixel, within the receptive field, that conforms to each pixel, within said projection grid, for which the value along the ρ axis is ρ0 using said correlation data, which conforms to the projection-direction angle θ, that has been stored in the first and second memories.

Claim 24

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto an axis (ρ axis) that perpendicularly intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; setting a projection grid, which has sides parallel to an a axis and the ρ axis, on the receptive field of the original two-dimensional image, in which the a axis is an axis in the projection direction; coordinate-transforming each pixel position within the projection grid, for which the value along the ρ axis is ρ0, into pixel positions within the receptive field; accumulating pixel data of each pixel, within the receptive field, obtained by the coordinate transformation; calculating, as projection values that prevail when the projection-direction angle is θ, accumulated values, at positions along the ρ axis, obtained by varying ρ0 incrementally a predetermined amount up to the width of the projection grid; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment; said coordinate transformation step including steps of: storing, in correlation with position along the ρ axis, a y address (where a pixel position within the receptive field is expressed by an x-y address) of a pixel group, within the receptive field, that corresponds to a pixel group of a=0 among pixels within said projection grid, and storing said correlation, in a first memory, for each and every projection direction; storing, in correlation with the y address, an x address of a pixel group, within the receptive field, that corresponds to a pixel group of ρ=0 among pixels within said projection grid, and storing said correlation, in a second memory, for each and every projection direction; and calculating an x, y address of each pixel, within the receptive field, that conforms to each pixel, within said projection grid, for which the value along the ρ axis is ρ0 using said correlation data, which conforms to the projection-direction angle θ, that has been stored in the first and second memories.

Claim 25

The computer-implemented method according to claim 20 wherein said step of calculating the x, y address includes steps of: obtaining, from said first memory, the y address (=y 0 ) of a pixel, within the receptive field, that corresponds to a first pixel at a position (0,ρ 0 ) within said projection grid; obtaining a value conforming to the y address (=y 0 ) from said second memory and adding ρ 0 to said value to obtain the x address; and subsequently incrementing the previous y address to obtain a y address of a pixel, within the receptive field, that corresponds to a second pixel, within the projection grid, for which the position along the ρ axis is ρ 0 , obtaining a value conforming the obtained address from said second memory, adding ρ 0 to said value to obtain the x address, and thereafter obtaining the x, y address in similar fashion.

Claim 26

A line-segment extracting computer-implemented method for subdividing an original image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto a ρ axis that intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; expressing a pixel position of the receptive field of the original two-dimensional image by an x-y address and making the ρ axis coincide with either the x axis or y axis; setting a projection grid on the receptive field, said projection grid defining a parallelogram having a side parallel to an a axis, which is the projection direction, and a side parallel to the ρ axis; subdividing said projection grid into slit-shaped parallelograms extending along the projection direction by a width of one pixel; calculating a projection value of the projection-direction angle θ at a position ρ0 along the ρ axis by accumulating image data of pixels, within the receptive field, that overlap a slit-shaped parallelogram corresponding to the position ρ0; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 27

A line-segment extracting computer-implemented method for subdividing an original image into receptive fields each of which defines a small area, projecting an image of the receptive field from a predetermined projection direction onto a ρ axis that intersects said projection direction to obtain projection values at positions along the ρ axis, subjecting the projection values to differentiation processing along the ρ axis, similarly subjecting projection values, which are obtained by successively varying a projection-direction angle θ incrementally a predetermined amount, to differentiation processing along the ρ axis to thereby obtain a differential value at each position on a ρ-θ plane, and extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the differential values, said method comprising the steps of: inputting an original two-dimensional image of an object; expressing a pixel position of the receptive field of the original two-dimensional image by an x-y address and making the ρ axis coincide with either the x axis or y axis; setting a projection grid on the receptive field, said projection grid defining a parallelogram having a side parallel to an a axis, which is the projection direction, and a side parallel to the ρ axis; subdividing said projection grid into slit-shaped parallelograms extending along the projection direction by a width of one pixel; calculating a projection value of the projection-direction angle θ at a position ρ0 along the ρ axis by accumulating image data of pixels, within the receptive field, that overlap a slit-shaped parallelogram corresponding to the position ρ0; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 28

The computer-implemented method according to claim 22, further including steps of: clipping a portion of an image of an enlarged receptive field, which has a range wider than that of the receptive field, from the original image and storing the clipped image in a memory; setting the projection grid, which is in the shape of the parallelogram having sides parallel to the a and ρ axes, on said enlarged receptive field and subdividing said projection grid into slit-shaped parallelograms extending along the projection direction by the width of one pixel; and calculating a projection value of the projection-direction angle θ at a position ρ 0 along the ρ axis by accumulating image data of pixels, within the enlarged receptive field, that overlap a slit-shaped parallelogram corresponding to the position ρ 0 .

Claim 29

The computer-implemented method according to claim 23, wherein when an enlarged receptive-field image conforming to the next adjoining receptive field is read out to the memory and image processing is performed after image processing regarding the present enlarged receptive-field image ends, a portion of the present enlarged receptive-field image that overlaps the next adjoining enlarged receptive-field image is left in the memory and only the portion of the next adjoining enlarged receptive-field image not overlapped is read out to the memory and subjected to image processing.

Claim 30

The computer-implemented method according to claim 23, wherein when an enlarged receptive-field image conforming to the next adjoining receptive field is read out to the memory and image processing is performed after image processing regarding the present enlarged receptive-field image ends, those projection values regarding the present enlarged receptive-field image that relate to an image portion in which the present enlarged receptive-field image and the next adjoining receptive-field image overlap are utilized as projection values of the next adjoining receptive-field image.

Claim 31

The computer-implemented method according to claim 22, wherein the width of each slit-shaped parallelogram in the direction that perpendicularly intersects the projection direction is made the width of one pixel.

Claim 32

The computer-implemented method according to claim 22, wherein if the width of each slit-shaped parallelogram in the direction that perpendicularly intersects the projection direction is less the width of one pixel, the adjacent projection value is used to apply a correction to the projection value of one pixel width in the direction that perpendicularly intersects the projection direction.

Claim 33

The computer-implemented method according to claim 22, wherein the step of accumulating image data of each pixel in the receptive field that overlaps each slit-shaped parallelogram includes steps of: storing, in a memory in correlation with a y address, an x address of a pixel group, within the receptive field, that corresponds to a pixel group constituting a slit-shaped parallelogram of ρ=0 among pixels within said projection grid, the correlation being stored in said memory as the projection-direction angle θ is changed; obtaining the y address of each pixel, within the receptive field, that corresponds to each pixel, within said projection grid, for which the value along the ρ axis is ρ 0 by successively incrementing the address y from an initial value, and obtaining an x address corresponding to said y address from said memory; obtaining an x address of each pixel within the receptive field by adding ρ 0 to said x address; and accumulating image data of each receptive-field pixel designated by each obtained x,y address, and calculating an accumulated value for which the value along the ρ axis is ρ 0 .

Claim 34

The computer-implemented method according to claim 22, wherein the step of accumulating image data of each pixel in the receptive field that overlaps each slit-shaped parallelogram includes steps of: storing an x address of a pixel, within the receptive field, that corresponds to a vertex pixel of ρ=0 within said projection grid; storing, in correlation with the y address, and for each and every projection direction, an incremental value dx of the x address of the pixel group, within the receptive field, that corresponds to the pixel group constituting the slit-shaped parallelogram of ρ 0 among the pixels within said projection grid; obtaining the y address of each pixel, within the receptive field, that corresponds to each pixel, within said projection grid, for which the value along the ρ axis is ρ 0 by successively incrementing the y address from an initial value; obtaining the x address of each receptive-field pixel by adding the incremental value dx of the x address conforming to said y address and the x address of said vertex point conforming to the projection direction θ; and calculating a projection value by accumulating image data of the receptive-field pixel designated by each obtained x-y address.

Claim 35

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of a receptive field from a predetermined projection direction, subjecting projection values to differentiation processing, obtaining differential values by successively varying a projection-direction angle θ incrementally a predetermined amount, extracting a line segment contained in the receptive-field image based upon a peak position of said differential values, and combining line segments of each of the receptive-field images, said method comprising the steps of: inputting an original two-dimensional image of an object; obtaining all projection values in a horizontal direction by projecting all receptive-field images of the original two-dimensional image along the horizontal direction from a prescribed projection direction, successively varying the projection-direction angle θ incrementally a predetermined amount and similarly storing projection values in a memory; successively reading the projection values of the projection-direction angle θ out of the memory and applying differentiation processing thereto; subsequently reading projection values, which prevail when the projection-direction angle θ is successively varied incrementally a predetermined amount, out of the memory and applying differentiation processing thereto; extracting line segments contained in the receptive-field images based upon peak positions of the differential values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segments.

Claim 36

A line-segment extracting computer-implemented method for subdividing an image into receptive fields each of which defines a small area, projecting an image of a receptive field from a predetermined projection direction, subjecting projection values to differentiation processing, obtaining differential values by successively varying a projection-direction angle θ incrementally a predetermined amount, extracting a line segment contained in the receptive-field image based upon a peak position of said differential values, and combining line segments of each of the receptive-field images, said method comprising the steps of: inputting an original two-dimensional image of an object; obtaining all projection values in a horizontal direction by projecting all receptive-field images of the original two-dimensional image along the horizontal direction from a prescribed projection direction, successively varying the projection-direction angle θ incrementally a predetermined amount and similarly storing projection values in a memory; successively reading the projection values of the projection-direction angle θ out of the memory and applying differentiation processing thereto; subsequently reading projection values, which prevail when the projection-direction angle θ is successively varied incrementally a predetermined amount, out of the memory and applying differentiation processing thereto; extracting line segments contained in the receptive-field images based upon peak positions of the differential values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segments.

Claim 37

The computer-implemented method according to claim 30, wherein said step of calculating all projection values in the horizontal direction includes steps of: providing m-number of data shifting means each having a variable amount of shift (where represents the number of pixels of the receptive field in the vertical direction thereof), and a memory for storing each amount of shift of the m-number of data shifting means in dependence upon the projection direction θ; successively inputting, to said m-number of data shifting means in linear fashion, image data of m-number of pixels in the vertical direction of all receptive-field images along the horizontal direction; causing each of said data shifting means to input, to adding means, a respective item of input image data upon shifting the image data by each amount of shift conforming to the projection direction angle θ; and causing said adding means to add m-number of items of image data outputted by said m-number of data shifting means.

Claim 38

A computer-implemented method of extracting a line segment contained in an original image, comprising the steps of: inputting an original two-dimensional image of an object; clipping an image, which conforms to the size of a receptive field defining a small area, from the original two-dimensional image; subjecting the receptive-field image that has been clipped to edge detection processing; projecting an image obtained by the edge detection processing from a prescribed projection direction upon an axis (ρ axis) that intersects said projection direction and obtaining a projection value at each position along the ρ axis; similarly obtaining projection values by successively varying the projection-direction angle θ incrementally a predetermined amount, and obtaining a projection value at each position on a ρθ plane; extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the projection values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 39

A computer-implemented method of extracting a line segment contained in an original image, comprising the steps of: inputting an original two-dimensional image of an object; clipping an image, which conforms to the size of a receptive field defining a small area, from the original two-dimensional image; subjecting the receptive-field image that has been clipped to edge detection processing; projecting an image obtained by the edge detection processing from a prescribed projection direction upon an axis (ρ axis) that intersects said projection direction and obtaining a projection value at each position along the ρ axis; similarly obtaining projection values by successively varying the projection-direction angle θ incrementally a predetermined amount, and obtaining a projection value at each position on a ρθ plane; extracting a line segment contained in the receptive-field image based upon a peak position (ρ,θ) of the projection values; and generating a one-dimensional image representing one of a contour line of the object and an edge between light and dark of the object based on the extracted line segment.

Claim 40

The computer-implemented method according to claim 32, further including subdividing the original image into receptive fields, applying the line-segment extraction processing to each receptive-field image and extracting a line segment contained in the original image.