JPH09145331A - Method and apparatus for recognizing object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an apparatus for recognizing the three-dimensional information of an object through image processing without touching the object at low cost in a short time. SOLUTION: A camera 2 picks up the image (first image) only of a belt conveyor 1, the image (second image) of belt conveyor 1 mounting a section paper 5 having a checkerboard pattern on the surface, and the image (fourth image) of belt conveyor 1 mounting an object A to be recognized. An image processor 42 produces an image (third image) indicative of difference between the first and second images, and an image (fifth image) indicative ofd the sum of third image and the difference between first and fourth images. Three- dimensional information (dimensions, direction and position) of the object A to be recognized is then recognized based on the fifth image (synthesized image of the object A and section paper 5).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object recognition method and device for recognizing three-dimensional information (dimension, direction, position) of an object in a non-contact manner, and more particularly to an image processing system using one camera. The present invention relates to an object recognition method and apparatus for recognizing these parameters of a 3D object.

[0002]

2. Description of the Related Art As a method of non-contactly recognizing a three-dimensional shape and a three-dimensional position of a three-dimensional object, a three-dimensional object to be recognized is imaged and obtained. A technique using image processing of an image is known. Two examples of a conventional method of recognizing an object in a three-dimensional space using such an image processing technique (a stereo image method and a slit light projection method) will be described below. These two examples are described in "3D Image Measurement" (published by: Shokodo, Seiji Iguchi,
Koto Sato).

FIG. 14 is a schematic diagram for explaining the principle of the stereo image method. Humans can obtain a stereoscopic view by looking at an object with both left and right eyes. This stereo imaging method
This is a method based on the concept that if there is a pair of images (stereo images) corresponding to the left and right eyes, it is possible to extract three-dimensional information according to these, and it is a passive method that observes the world as it is like human eyes Is a typical method. As shown in FIG. 14, two cameras such as an ITV camera are arranged on the left and right, and three-dimensional information is recognized from the obtained image information by the triangulation method.

In order to extract three-dimensional information from such a stereo image, it is necessary to detect which point on one screen corresponds to which point on the other screen. This is the corresponding point determination (matching) in the stereo image method
However, there is a problem that a technique for sufficiently accurately detecting the corresponding point has not been established, and as a result, only uncertain three-dimensional information can be obtained.

FIG. 15 is a schematic diagram for explaining the principle of the slit light projection method (light cutting method). In this slit light projection method, as shown in FIG. 15, slit light is projected on an object to be recognized, a slit projected image is taken by a TV camera from a direction orthogonal to this, and the image is processed by an image processing device. This is a method of recognizing the three-dimensional information of the object.

However, this method requires a slit light source device, and since a large amount of data must be handled, there is a problem that it takes a long processing time.

The present invention has been made in view of the above circumstances, and provides an object recognition method and apparatus which can reduce the cost as compared with a stereo image method using an image processing system using two cameras. With the goal.

Another object of the present invention is to provide a method and apparatus for recognizing an object which can increase the processing speed as compared with the slit light projection method.

[0009]

An object recognition method according to claim 1 is a method of recognizing three-dimensional information of an object by using image processing of an image of the object, wherein an object to be recognized is arranged. A step of capturing only the recognition environment to obtain the first image, a step of disposing a jig having a grid-like line in the recognition environment, and an image of the recognition environment in which the jig is arranged are imaged. A step of obtaining a second image, a step of obtaining a third image which is a difference between the second image and the first image, a step of arranging an object to be recognized in the recognition environment, and a state of arranging the object. Based on the fifth image, a step of capturing the recognition environment to obtain a fourth image, a step of obtaining a fifth image of a sum of the difference between the fourth image and the first image and the third image, Calculating the three-dimensional information of the object to be recognized.

An object recognition method according to a second aspect of the present invention is an object recognition method that uses the
In a method of recognizing dimensional information by using image processing of an image of an image of the object, a step of imaging only a recognition environment in which an object to be recognized is arranged to obtain an eleventh image, and grid lines are shown. A step of arranging the jig and the object to be recognized in the recognition environment; a step of imaging the recognition environment with the jig and the object to be recognized to obtain a twelfth image; It is characterized by including a step of obtaining a thirteenth image which is a difference between the image and the eleventh image, and a step of calculating three-dimensional information of the object to be recognized based on the thirteenth image.

An object recognition apparatus according to claim 3 is the object recognition device according to claim 1.
An apparatus used for carrying out the described method, comprising image processing means for obtaining the third image and fifth image, and computing means for computing three-dimensional information of the object to be recognized based on the fifth image. It is characterized by including.

According to a fourth aspect of the object recognition apparatus, the object recognition apparatus
In an apparatus for recognizing dimensional information using image processing of an image of an image of the object, there is nothing in the recognition environment in which an image pickup means, a jig with a grid line, and an object to be recognized are to be arranged. First image processing means for obtaining a first edited image of a difference between an image in a first state that is not arranged and an image in a second state in which the jig is arranged in the recognition environment, and a recognition target of the recognition environment in the recognition environment. Second image processing means for obtaining an image in a third state in which an object is arranged and a second edited image of a difference between the images in the first state;
A third image processing means for obtaining a third edited image that is the sum of the edited image and the second edited image; and a computing means for computing three-dimensional information of the object to be recognized based on the third edited image. Is characterized by.

The object recognition apparatus according to claim 5 is the object recognition device according to claim 2.
An apparatus used for carrying out the method as described, comprising: image processing means for obtaining the thirteenth image; and computing means for computing three-dimensional information of the object to be recognized based on the thirteenth image. Characterize.

According to a sixth aspect of the object recognition apparatus, the object recognition apparatus
In an apparatus for recognizing dimensional information using image processing of an image of an image of the object, there is nothing in the recognition environment in which an image pickup means, a jig with a grid line, and an object to be recognized are to be arranged. Based on the edited image, an image processing means for obtaining an edited image of the difference between the image in the 11th state that is not arranged, and the image in the 12th state where the jig and the object to be recognized are arranged in the recognition environment. And a calculating unit that calculates three-dimensional information of the object to be recognized.

The outline of the three-dimensional object recognition processing according to claims 1, 3 and 4 will be described. The camera is in each of the state of only the recognition environment in which the three-dimensional object recognition process is performed, the state in which a jig with grid lines is placed in the recognition environment, and the state in which the object to be recognized is placed in the recognition environment. The image of each state is acquired as a first image, a second image, and a fourth image. The order of acquiring these images may be arbitrary. Then, an image (third image) showing the difference between the first image and the second image is obtained by image processing. The third image is an image of only the jig. 4th image and 1st
An image showing the difference from the image (image of only the object to be recognized) and a fifth image showing the sum of the third images are obtained by image processing. The fifth image is a composite image of the object to be recognized and the jig. Grid lines for specifying the coordinates are shown on the jig, and the composite image (fifth image) shows the grid lines and the object to be recognized in an overlapping manner. 5
Three-dimensional information (dimension, direction, position) of the object to be recognized can be calculated from the image.

The outline of the three-dimensional object recognition processing according to claims 2, 5, and 6 will be described. Images of each state are taken by the camera for each of the state of the recognition environment in which the three-dimensional object recognition process is performed, the jig in which the grid lines are shown, and the state of the object to be recognized placed in the recognition environment. Are acquired as the 11th image and the 12th image. The order of acquiring these images may be arbitrary. Then, an image (13th image) showing the difference between the 11th image and the 12th image is obtained by image processing. This thirteenth image is a composite image of the object to be recognized and the jig. Grid lines for identifying the coordinates are shown on the jig, and the composite image (13th image) shows the grid lines and the object to be recognized in an overlapping manner. Three-dimensional information (dimension,
Direction, position) can be calculated.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below with reference to the drawings showing the embodiments.

In the embodiments described below, the known information that the object to be recognized is a rectangular parallelepiped is combined with the two-dimensional information obtained from the monocular image processing apparatus to combine the three-dimensional information of the object (dimension, direction and position of the object). ) Will be described. FIG. 1 is a schematic diagram showing an example of an implementation state of the object recognition method of the present invention.

In FIG. 1, reference numeral 1 is a belt conveyor for conveying a recognition target object A to be recognized. In the vicinity of the belt conveyer 1, a camera 2 for taking an image of the object A to be recognized, the belt conveyer 1 and the like to obtain an image as described later is arranged, and illuminations 3 and 3 for illuminating the image pickup position of the camera 2. Is provided. Then, the recognized object A is placed and conveyed on the belt conveyor 1 in any direction,
The image is obtained by temporarily stopping within the visual area of the camera 2. The field of view of camera 2 is 512 x 48
It has 0 pixels and has a general resolution.

The image signal obtained by the camera 2 is sent to the image processing section 4. The image processing unit 4 has a camera input I / F (interface) 41 for converting an image signal input from the camera 2 into processable image information, and an image processing processor 42 for performing arithmetic processing on the image information. It has an image memory 43 for storing image information and a personal computer 44 including a keyboard for external input.

Numeral 5 is a graph paper as a jig having a grid-like line on its surface. The jig may be made of any material as long as it can be easily placed on the belt conveyor 1, and it is needless to say that a material other than the graph paper may be used.

Next, the operation will be described. Figures 2 and 3
3 is a flowchart showing a processing procedure of an object recognition method of the present invention.

First, an image of a recognition environment (this is defined as a first image) to be a place for recognizing a three-dimensional object is obtained (step S1). Specifically, the camera 2 acquires an image of the belt conveyor 1 in which nothing is placed, and the image information is stored in the image memory 43. FIG. 4 schematically shows this first image. As shown in FIG. 4, the coordinate axes Xc and Yc of the camera coordinate system are defined, and the intersection of both coordinate axes (the origin of the camera coordinate system) is defined as the camera coordinate origin Oc. Set.

Next, the jig with the lines in a grid pattern is placed in the recognition environment (step S2), and an image of the placed state (this is defined as the second image) is obtained (step S).
3). Specifically, the graph paper 5 is placed on the belt conveyor 1, the image of the state is acquired by the camera 2, and the image information is stored in the image memory 43. FIG. 5 schematically shows this second image. Since the camera 2 photographs the jig (graph paper 5) from an oblique direction, the shape of the grid of the jig is recognized as a grid approximated by a parallelogram. Further, the coordinate axes Xb and Yb of the jig coordinate system (base coordinate system) are defined, and the intersection of the two coordinate axes (the origin of the base coordinate system) is set as the camera position Ob.

Each cross point of the grid in the jig is defined as P _{i, j} . However, i represents a row number and j represents a column number. As the coordinate value of P _{i, j} , there are a base coordinate value in the base coordinate system and a camera coordinate value in the camera coordinate system. Therefore, the contents of coordinates of P _{i, j} are as shown in (1) below. P _{i, j} = (Xb _{i, j} , Yb _{i, j} , Xc _{i, j} , Yc _{i, j} ) (1)

Since the position of the jig with respect to the camera can be set arbitrarily, each intersection P _{i, j of} the jig seen from the camera
The base coordinate values Xb _{i, j} and Yb _{i, j of} are known. Also, the camera coordinate values Xc _{i, j} , Yc of each intersection P _{i, j
i and j} are also known because they can be directly determined from the two-dimensional image data from the camera. The image processor 42 calculates the base coordinate value and the camera coordinate value at each intersection P _{i, j} .
In the internal memory (not shown) of the device (step S
4). Further, as shown in FIG. 6, the distance from the base coordinate system plane to the camera 2 is defined as the camera height H.

Next, an image (defined as a third image) obtained by calculating the difference between the second image and the first image is obtained (step S).
5). Specifically, the image information of the belt conveyor 1 on which the graph paper 5 is placed and the image information of only the belt conveyor 1 are read from the image memory 43 to the image processor 42, and the difference between the two image information is calculated. Then, the image information indicating the difference is stored in the image memory 43. FIG. 7 schematically shows this third image, and only the grid-like lines of the jig (graph paper 5) are extracted.

Then, a thinning process is performed (step S
6). In image processing, when grayscale data is differentiated and binarized, a portion to be shown by a line may become an elongated surface. The process for converting this elongated surface into a line is a thinning process.

The jig is removed from the recognition environment and the object to be recognized is placed in the recognition environment (step S7). Specifically, after removing the graph paper 5 from the belt conveyor 1, the rectangular parallelepiped object A to be recognized is placed on the belt conveyor 1 and conveyed.
Stop in the visual area of the camera 2. Then, an image in which the object to be recognized is placed in the recognition environment (this is defined as the fourth image) is obtained (step S8). Specifically, the image of the object A to be recognized placed on the belt conveyor 1 is acquired by the camera 2, and the image information is stored in the image memory 43. FIG. 8 schematically shows this fourth image.

When the difference between the fourth image and the first image is calculated,
Only the object to be recognized can be extracted (step S9). FIG. 9 shows an image in which only this recognized object (rectangular solid) is extracted. As shown in FIG. 9, the sides of the extracted rectangular parallelepiped closest to the camera are defined as x and y, and the end points of the side x are q ₁ , q ₂ ,
The end points of the side y are defined as q ₂ and q _3, and the intersection point q of both sides x and y is defined.
_The remaining side passing through ₂ is defined as z. Further, a point where a straight line connecting the remaining end points of the side z and the camera intersects the plane defined by both sides x and y is defined as q ₄ .

After subjecting the image obtained by extracting only the object to thinning processing (step S10), four points q _{1 of the} object to be recognized,
The camera coordinate values in the camera coordinate system of q ₂ , q ₃ , and q ₄ are stored in the internal memory of the image processing processor 42 (step S11).

Next, an image (defined as a fifth image) obtained by calculating the sum of the image obtained by extracting only the object to be recognized and the third image is obtained (step S12). FIG. 10 schematically shows the fifth image, which is an image in which the object to be recognized is placed on the grid-shaped line of the jig.

Here, if the base coordinate values of the four points q ₁ , q ₂ , q ₃ , and q ₄ can be calculated, each side, position, and direction of the rectangular parallelepiped which is the object to be recognized are obtained based on these coordinate values. be able to. Hereinafter, this calculation method will be described.

First, a method of obtaining the base coordinate value of the point q ₃ (steps S13 to S19) will be described. The point q ₃ is
As shown in FIG. 10, four cross points P _2,6 , P
It exists inside a quadrangle surrounded by _1,6 , P _1,7 and P _2,7 . FIG. 11 shows an enlarged view near the point q ₃ . In FIG.
The Xb1 axis and the Yb6 axis correspond to the Xb axis and the Yb axis when the origin Ob is translated to the point P _1,6 . 4 points P _1,6 ,
The camera coordinate values and base coordinate values of P _1,7 , P _2,6 and P _2,7 are read from the memory (step S13), and the following calculation is performed.

First, the straight line P _1,6 P _1,7 and the straight line P _2,6 P
It is determined whether or not ₂ and ₇ are parallel (step S1
Four). The opposite sides of the quadrangle P _2,6 P _1,6 P _1,7 P _2,7 are generally not parallel due to the effect of image distortion due to monocular vision. When they are not parallel to each other, the opposite sides P _2,6 P _1,6 and P _2,7 P _1,7 are respectively extended to make the intersection point O _1, and the opposite side P 2
_2,6 P _2,7 and P _1,6 P _1,7 are respectively extended and the intersection point is O ₂ .

The intersection O of these points₁, O_TwoCamera
Calculate the camera coordinate values in the coordinate system (step S1
Five). 4 points P_1,6, P_1,7, P_2,6, P_2,7Camera seat
Since the standard value is known, the point O₁, O_TwoThe camera coordinate value of is
Can be calculated. Specifically, the straight line P_2,6P_1,6And the straight line P
_2,7P_1,7The equation is calculated as
O ₁Calculate the camera coordinate value of_2,6P
_2,7And the straight line P_1,6P_1,7And the equation of
Point O as coordinates_TwoCalculate the camera coordinate value of.

The point O₁And the point q_ThreeConsider a straight line connecting
Eh, this straight line and side P_1,6P_1,7, Side P _2,6P_2,7Interaction with
Each point is r₁, R_ThreeAnd Similarly, the point O_TwoAnd the point q
_ThreeConsider a straight line connecting to and this line and side P_2,7P_1,7， Side
P_2,6P_1,6Each intersection with_Two, R_FourAnd here
And the point q_ThreeSince the camera coordinate value of is known, each intersection r
₁, R_Two, R_Three, R_FourThe camera coordinate value of can be calculated (
Step S16). Specifically, straight line O_Twoq_ThreeThe formula of
This straight line O_Twoq_ThreeAnd the straight line P already obtained_2,6P_1,6Interaction with
Point r as coordinates of point_FourWhen calculating the camera coordinate values of
And a straight line O₁q_ThreeThen, the straight line O₁q_ThreeAnd already
Obtained straight line P_1,6P_1,7Point r as the coordinate of the intersection with₁of
Calculate camera coordinate values.

Then, the calculated point r_FourTo the camera coordinate value of
Therefore, the point q_ThreeIn the Yb axis direction
A base coordinate value is calculated (step S17). In Figure 12,
Line P _2,6P_1,6The enlarged view of FIG. Point P on this straight line
_1,6, P_2,6, R_FourThe contents of each coordinate of
Is defined as P_1,6= (Xb_1,6, Yb_1,6, Xc_1,6, Yc_1,6) P_2,6= (Xb_2,6, Yb_2,6, Xc_2,6, Yc_2,6) R_Four= (Xb_Four, Yb_Four, Xc_Four, Yc_Four) Where point P_1,6, P_2,6Is the camera coordinate value (Xc_1,6,
Yc_1,6), (Xc_2,6, Yc_2,6) And base coordinate values
(Xb_1,6, Yb_1,6), (Xb_2,6, Yb_2,6) Is already
Is knowledge. Also, the point r_FourIs the camera coordinate value (Xc_Four, Yc
_Four) Is determined by the above calculation method, but the base coordinates
Value (Xb_Four, Yb_Four) Is unknown at this stage.

In FIG. 12, the lengths of the line segment P _2,6 r ₄ and the line segment P _2,6 P _1,6 in the camera coordinate system are respectively d.
Assuming c ₁ and dc ₂ , they are as follows. dc ₁ = {(Xc _2,6 -Xc ₄ ) ² + (Yc _2,6- Yc ₄ ) ² } ^1/2 dc ₂ = {(Xc _2,6- Xc _1,6 ) ² + (Yc _{2, 6-} Yc _1,6 ) ² } ^1/2

By the way, the base coordinate values of the points P _1,6 , the points P _2,6 and the point r ₄ in the Xb axis direction are equal by the definition of the base coordinate system. Also, the base coordinate values Yb _1,6 and Yb _2,6 of the points P _1,6 and P _2,6 are known, and only the base coordinate value Yb ₄ of the point r ₄ is unknown. here,
When the distance between the lines of the grid of the jig is sufficiently small, the following proportional expression is established. _{dc 1: dc 2 = | Yb} 2,6 -Yb 4 |: | Yb 2,6 -Yb 1,6 | by substituting the calculated values of dc _1, dc ₂ described above in the proportional, the point r ₄ The base coordinate value Yb ₄ of can be obtained. If the distance between the grids of the jig is sufficiently small, it can be approximated that the point r ₄ and the point q ₃ have the same base coordinate value in the Yb axis direction. Therefore, the base of the point q _{3 in} the Yb axis direction can be approximated. The coordinate value can be obtained.

If the straight line P _1,6 P _1,7 and the straight line P _2,6 P _2,7 are parallel (step S 14: YES), the point q ₃
And the point r ₄ have the same base coordinate value in the Yb axis direction, the base coordinate value (Yb _q3 ) in the Yb axis direction of the point q ₃ can be obtained based on the following proportional expression (step S18). ). _{dc 1: dc 2 = | Yb} 2,6 -Yb q3 |: | Yb 2,6 -Yb 1,6 |

Further, in the same manner as the method of calculating the base coordinate value in the Yb-axis direction described above, the base coordinate value in the Xb-axis direction of the point q ₃ is calculated from the calculated camera coordinate value of the point r ₄ (step S19). ). Therefore, the base coordinate value of the point q ₃ can be obtained by the above.

Then, the base coordinate values of the points q ₁ , q ₂ and q ₄ are obtained by the same method as the method of calculating the base coordinate value of the point q ₃ described above (step S20).

A side x of the base coordinate system is obtained from the base coordinate values of the points q ₁ and q ₂ , and a side y of the base coordinate system is obtained from the base coordinate values of the points q ₂ and q ₃ (step S21). Further, the orientation of the object to be recognized (rectangular solid) can be obtained from the inclination between the side x and the Xb axis of the base coordinate system.

Next, the side z of the base coordinate system is obtained from the base coordinate values of the points q ₂ and q ₄ (step S22). Figure
FIG. 13 is a schematic diagram showing the principle of this determination method. In FIG. 13, q ₄ ′ is the other end point of the side z, and the distances from the camera position to the points q ₂ and q ₄ are A ₂ and A ₄ , respectively. The distances A ₂ and A ₄ can be calculated from the base coordinate values of the points q ₂ and q ₄ . The side z can be calculated based on the following proportional expression. _{H: z = A 4: (} A 4 -A 2)

As described above, it is possible to recognize the absolute position, orientation, and size of the object to be recognized having a rectangular parallelepiped shape by using the image processing system using a single camera.

When the recognition process for one object to be recognized is completed, it is judged whether or not the object to be recognized remains to be recognized (step S23). If it remains, the process returns to step S7, and the object to be recognized is placed in the recognition environment again (the next object to be recognized A is placed on the belt conveyor 1).
The processing operation is repeated from.

In the above example, the case where the object to be recognized has a rectangular parallelepiped shape has been described. However, recognition processing can be similarly performed on objects to be recognized having other shapes such as a triangular prism, a quadrangular pyramid, and a triangular pyramid. Needless to say.

In the above-described embodiment, the first image in the state where only the belt conveyor 1 is used, the graph paper 5 is used.
The second image of the state in which the object A to be recognized is placed on the belt conveyor 1 and the fourth image in the state of being placed on the belt conveyor 1 are acquired by the camera 2 in this order, but it is sufficient that the fifth image is finally obtained. , Each of these images may be acquired in any order.

Further, in the above-described embodiment, the case where the recognition environment is on the belt conveyor 1 is assumed, so that the recognition target object A is removed after removing the graph paper 5 (jig) so as to be practical. I try to put it on. Of course, the recognized object may be placed directly on the jig placed in the recognition environment. The processing procedure in such a case will be described below.

First, an image of the recognition environment in which nothing is placed (this is defined as the 11th image) is acquired by the camera. This eleventh image is the same as the above-mentioned first image (see FIG. 4). Next, the jig and the object to be recognized are placed in the recognition environment, and an image of the recognition environment with these placed (this is defined as a twelfth image) is acquired by the camera. An image showing the difference between the 12th image and the 11th image.
(Defined as an image) is obtained by image calculation processing. This thirteenth
The image is the same as the above-mentioned fifth image (see FIG. 10). Then, based on the obtained thirteenth image, three-dimensional information (dimension, direction and position) of the recognized object is obtained. In addition,
The calculation method for obtaining the three-dimensional information in this case is the same as the above-described calculation method based on the fifth image, and thus description thereof will be omitted.

[0052]

As described above, since only one camera needs to be installed, it is less expensive than the stereo image method using two cameras, and the known information on the grid of the jig is used. , The recognition processing speed is faster than the slit light projection method, and since the image showing the difference between the image of the recognition target object placed in the recognition environment and the image of the recognition environment only is used, the recognition environment is The present invention has excellent effects such as being able to detect three-dimensional information of an object to be recognized even if it is complicated.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing an implementation state of an object recognition method of the present invention.

FIG. 2 is a flowchart showing a procedure of an object recognition method of the present invention.

FIG. 3 is a flowchart showing a procedure of an object recognition method of the present invention.

FIG. 4 is a schematic diagram of a first image showing only a recognition environment.

FIG. 5 is a schematic diagram of a second image showing a state in which a jig is placed in a recognition environment.

FIG. 6 is a diagram for explaining the definition of a camera height H.

FIG. 7 is a third image (= second image-first image) showing the jig.
FIG.

FIG. 8 is a schematic diagram of a fourth image showing a state in which an object to be recognized is placed in a recognition environment.

FIG. 9 is a schematic diagram showing an object to be recognized.

FIG. 10 is a fifth image (= fourth image) showing a recognized object and a jig.
It is a schematic diagram of (image-first image + third image).

FIG. 11 is an enlarged view of the vicinity of one vertex of a recognized object.

FIG. 12 is a further enlarged view of FIG. 11.

FIG. 13 is a diagram for explaining the principle of obtaining one side of a recognized object.

FIG. 14 is a schematic diagram for explaining the principle of a conventional method (stereo image method).

FIG. 15 is a schematic diagram for explaining the principle of a conventional method (slit light projection method).

[Explanation of symbols]

 1 Belt Conveyor 2 Camera 3 Illumination 4 Image Processor 5 Graph Paper (Jig) 42 Image Processor 43 Image Memory A Object to be recognized

Claims (6)

[Claims] 1. A method of recognizing three-dimensional information of an object using image processing of an image of the object, the method comprising: capturing only a recognition environment in which an object to be recognized is arranged to obtain a first image. A step of arranging a jig having a grid-like line in the recognition environment, a step of capturing an image of the recognition environment with the jig arranged to obtain a second image, the second image and the A step of obtaining a third image which is a difference from the first image, a step of arranging an object to be recognized in the recognition environment, and a step of imaging the recognition environment with the object arranged to obtain a fourth image A step of obtaining a fifth image of a sum of the difference between the fourth image and the first image and the third image, and a step of calculating three-dimensional information of the object to be recognized based on the fifth image. An object recognition method comprising: 2. A method of recognizing three-dimensional information of an object by using image processing of an image of the object, the method comprising the step of imaging only a recognition environment in which an object to be recognized is arranged to obtain an eleventh image. A step of arranging the jig and the object to be recognized in which the grid-like lines are shown in the recognition environment, and an image of the recognition environment in which the jig and the object to be recognized are arranged.
A step of obtaining 12 images, a step of obtaining a 13th image which is a difference between the 12th image and the 11th image, and a step of calculating three-dimensional information of the recognition target object based on the 13th image. An object recognition method having. 3. An apparatus used for carrying out the method according to claim 1, comprising image processing means for obtaining the third image and the fifth image, and the object to be recognized 3 based on the fifth image. An object recognition apparatus, comprising: a calculation unit that calculates dimension information. 4. An apparatus for recognizing three-dimensional information of an object by using image processing of an image of the object, the imaging means, a jig having a grid line, and an object to be recognized. An image of the first state where nothing is placed in the recognition environment to be placed,
And first image processing means for obtaining a first edited image of a difference between images in a second state in which the jig is placed in the recognition environment, and an image in a third state in which an object to be recognized is placed in the recognition environment, And a second obtaining an edited image of the difference between the images in the first state.
Image processing means, third image processing means for obtaining a third edited image that is the sum of the first edited image and the second edited image, and three-dimensional information of the object to be recognized based on the third edited image. An object recognition device, comprising: a calculation unit that calculates. 5. An apparatus used for carrying out the method according to claim 2, wherein image processing means for obtaining the thirteenth image and three-dimensional information of the object to be recognized are calculated based on the thirteenth image. An object recognition apparatus comprising: 6. An apparatus for recognizing three-dimensional information of an object by using image processing of an image of the object, the imaging means, a jig having a grid line, and an object to be recognized. An image of the 11th state where nothing is placed in the recognition environment that should be placed,
And image processing means for obtaining an edited image of the difference between the images in the twelfth state in which the jig and the object to be recognized are arranged in the recognition environment, and three-dimensional information of the object to be recognized based on the edited image. An object recognition device, comprising: a calculation unit that calculates.