System and method for recognition of image information

Claim 2

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: edge detecting means for detecting an edge component from input digital image data; first binarization processor means for processing an image obtained by the edge detecting means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; and linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means.

Claim 3

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data and for subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; first binarization processor means for processing the difference image data obtained by the mean value calculation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; and linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means.

Claim 4

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data and for subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; first binarization processor means for processing the difference image data obtained by the mean value calculation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; and linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means.

Claim 5

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data; modified Hough transformation means for applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and for effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the mean value M calculated by the mean value calculation means; binarization processor means for processing the modified Hough transformed image from the modified Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; and linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means.

Claim 6

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data; modified Hough transformation means for applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and for effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the mean value M calculated by the mean value calculation means; binarization processor means for processing the modified Hough transformed image from the modified Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; and linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means.

Claim 7

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: edge detecting means for detecting an edge component from input digital image data; first binarization processor means for processing an image obtained by the edge detecting means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 8

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: edge detecting means for detecting an edge component from input digital image data; first binarization processor means for processing an image obtained by the edge detecting means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 9

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data and for subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; first binarization processor means for processing the difference image data obtained by the mean value calculation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 10

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data and for subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; first binarization processor means for processing the difference image data obtained by the mean value calculation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; Hough transformation means for applying a Hough transformation to the binarization image obtained by the first binarization processor means; second binarization processor means for processing the Hough transformed image from the Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 11

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data; modified Hough transformation means for applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and for effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the mean value M calculated by the mean value calculation means; binarization processor means for processing the modified Hough transformed image from the modified Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 12

An image information recognition system for inspecting, recognizing or judging an object using an image thereof, comprising: mean value calculation means for calculating a mean value M of all picture element values from input digital image data; modified Hough transformation means for applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and for effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the mean value M calculated by the mean value calculation means; binarization processor means for processing the modified Hough transformed image from the modified Hough transformation means by use of a threshold to provide binarization image data comprising active picture elements and inactive picture elements; labeling processor means for grouping adjacent picture elements in the binarization image obtained by the second binarization processor means; representative point calculation means for determining the coordinates of a representative point of each group of adjacent picture elements grouped by the labeling processor means; linear component estimation means for determining a linear equation for a rectilinear portion in the edge component in the original image from the coordinate information of respective representative points found by the representative point calculation means; spacing calculation means operative when there are a plurality of rectilinear portions oriented in the same direction as determined by the linear component estimation means to calculate a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and minimum value calculation means for determining the minimum spacing among the spacings calculated by the spacing calculation means and for outputting it as representative of the width of an object to be recognized in the original image.

Claim 13

An image information recognition system according to one of claims 4 to 6 , wherein the minimum value calculation means is replaced by equal spacing decision means which determines the minimum spacing among a plurality of rectilinear portions having the same inclination, compares it with the remaining spacings, and in the event the same spacing occurs at a plurality of occurrences, that spacing is outputted as representative of the width of equally spaced linear components throughout the entire image.

Claim 14

An image information recognition system according to one of claims 4 to 6 , wherein the minimum value calculation means is replaced by equal spacing decision means which determines the minimum spacing among a plurality of rectilinear portions having the same inclination, compares it with the remaining spacings, in the event the same spacing does not occur at a plurality of locations, determines a second minimum spacing for comparison with the remaining spacings, repeating such procedure until the same spacing occurs at a plurality of locations, whereupon that spacing is outputted as representative of the width of equally spaced linear components throughout the entire image, and in the event the same spacing does not occur at a plurality of locations, outputs the minimum spacing which is initially determined as representing the width of the object to be recognized.

Claim 15

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: detecting an edge component from input digital image data; applying a first threshold operation to the image obtained by the detection of the edge component to provide a binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained as a result of the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 16

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: detecting an edge component from input digital image data; applying a first threshold operation to the image obtained by the detection of the edge component to provide a binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained as a result of the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 17

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data, and subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; applying a first threshold operation to the difference image data to provide binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained by the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 18

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data, and subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; applying a first threshold operation to the difference image data to provide binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained by the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 19

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data; applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the calculated mean value M; applying a threshold operation to the image obtained by the polling to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 20

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data; applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the calculated mean value M; applying a threshold operation to the image obtained by the polling to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; and determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found.

Claim 21

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: detecting an edge component from input digital image data; applying a first threshold operation to the image obtained by the detection of the edge component to provide a binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained as a result of the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 22

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: detecting an edge component from input digital image data; applying a first threshold operation to the image obtained by the detection of the edge component to provide a binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained as a result of the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 23

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data, and subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; applying a first threshold operation to the difference image data to provide binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained by the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 24

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data, and subtracting the mean value M from the picture element value of each of the picture elements in the original image data to produce a difference image data; applying a first threshold operation to the difference image data to provide binarization image data comprising active picture elements and inactive picture elements; applying a Hough transformation to the binarization image obtained by the first threshold operation; applying a second threshold operation to the Hough transformed image to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation; determining the coordinates of a representative point of each group of adjacent picture elements each of which have been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 25

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data; applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the calculated mean value M; applying a threshold operation to the image obtained by the polling to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation together; determining the coordinates of a representative point of each of a plurality of labels each of which has been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 26

An image information recognition method for inspecting, recognizing or judging an object using an image thereof, comprising the steps of: calculating a mean value M of all picture element values from input digital image data; applying a Hough transformation ρ=x cos θ+y sin θ to each picture element (its coordinates (x, y)) of the original image and effecting a polling in the θ−ρ parameter space with a weight equal to the difference |t(x, y)−M| between the picture element value t and the calculated mean value M; applying a threshold operation to the image obtained by the polling to provide binarization image data comprising active picture elements and inactive picture elements; grouping adjacent picture elements in the binarization image obtained by the second threshold operation together; determining the coordinates of a representative point of each of a plurality of labels each of which has been grouped; determining a linear equation for a rectilinear portion of the edge component in the original image from coordinate information of respective representative points found; in the event there are a plurality of rectilinear portions oriented in the same direction, calculating a spacing between adjacent rectilinear portions from information representing intercepts of the plurality of rectilinear portions; and determining the minimum spacing among the calculated spacings and determining it as representative of the width of the object to be recognized in the original image.

Claim 27

An image information recognition method according to one of claims 12 to 14 , wherein instead of the step of determining the minimum value spacing among the calculated spacings and determining it to be representative of the width of the object to be recognized in the original image, is used the step of comparing the minimum spacing among a plurality of rectilinear portions having the same inclination with the remaining spacings, and if the same spacing occurs at a plurality of locations, determining that spacing to be representative of the width between two adjacent linear components in equally spaced linear components throughout the entire image.

Claim 28

An image information recognition method according to one of claims 12 to 14 , wherein instead of the step of determining the minimum spacing among the calculated spacings and determining it to be representative of the width of the object to be recognized in the original image, is used the step of comparing the minimum spacing among a plurality of rectilinear portions having the same inclinations with the remaining spacings, in the event the same spacing does not occur at a plurality of locations, determining a second minimum spacing and comparing with the remaining spacings, and repeating such procedure until the same spacing occurs at a plurality of locations, whereupon such spacing is outputted as representative of the width between two adjacent linear components in equally spaced linear components throughout the entire image, and in the event the same spacing does not occur at a plurality of locations ultimately, determining the minimum spacing which was initially determined to be representative of the width of the object to be recognized.